begin with curl

fix building mm::structure using pdb_seq_num instead of auth_seq_num

.github/workflows/build-documentation.yml

@@ -19,7 +19,7 @@ jobs:
   docs:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v1
+    - uses: actions/checkout@v4
 
     - name: Set reusable strings
       # Turn repeated input strings (such as the build output directory) into step outputs. These step outputs can be used throughout the workflow file.
@@ -47,7 +47,7 @@ jobs:
         ls -l ${{ steps.strings.outputs.build-output-dir }}/docs/sphinx
 
     - name: Upload artifact
-      uses: actions/upload-pages-artifact@v2
+      uses: actions/upload-pages-artifact@v3
       with:
         path: ${{ steps.strings.outputs.build-output-dir }}/docs/sphinx
 
@@ -62,4 +62,4 @@ jobs:
     steps:
       - name: Deploy to GitHub Pages
         id: deployment
-        uses: actions/deploy-pages@v2
+        uses: actions/deploy-pages@v4

CMakeLists.txt

@@ -24,10 +24,15 @@
 
 cmake_minimum_required(VERSION 3.23)
 
+if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR AND NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
+	set(CMAKE_BUILD_TYPE Release CACHE STRING "Build type" FORCE)
+	set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS "Debug" "Release" "MinSizeRel" "RelWithDebInfo")
+endif()
+
 # set the project name
 project(
 	libcifpp
-	VERSION 8.0.0
+	VERSION 9.0.0
 	LANGUAGES CXX)
 
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
@@ -63,25 +68,26 @@ elseif(MSVC)
 endif()
 
 # Build documentation?
-option(BUILD_DOCUMENTATION "Build the documentation" OFF)
+set(BUILD_DOCUMENTATION OFF CACHE BOOL "Build the documentation")
 
 # Optionally build a version to be installed inside CCP4
-option(BUILD_FOR_CCP4 "Build a version to be installed in CCP4")
+set(BUILD_FOR_CCP4 OFF CACHE BOOL "Build a version to be installed in CCP4")
+
+# Optionally use libcurl to fetch compound files
+set(USE_CURL_FOR_CCD ON CACHE BOOL "Use curl to fetch missing CCD files")
 
 # Building shared libraries?
 if(NOT(BUILD_FOR_CCP4 AND WIN32))
-	option(BUILD_SHARED_LIBS "Build a shared library instead of a static one" OFF)
+	set(BUILD_SHARED_LIBS OFF CACHE BOOL "Build a shared library instead of a static one")
 endif()
 
 if(PROJECT_IS_TOP_LEVEL AND NOT BUILD_FOR_CCP4)
 	# Lots of code depend on the availability of the components.cif file
-	option(CIFPP_DOWNLOAD_CCD
-		"Download the CCD file components.cif during installation" ON)
+	set(CIFPP_DOWNLOAD_CCD ON CACHE BOOL "Download the CCD file components.cif during installation")
 
 	# An optional cron script can be installed to keep the data files up-to-date
 	if(UNIX AND NOT APPLE)
-		option(CIFPP_INSTALL_UPDATE_SCRIPT
-			"Install the script to update CCD and dictionary files" ON)
+		set(CIFPP_INSTALL_UPDATE_SCRIPT ON CACHE BOOL "Install the script to update CCD and dictionary files")
 	endif()
 else()
 	unset(CIFPP_DOWNLOAD_CCD)
@@ -91,14 +97,13 @@ endif()
 # When CCP4 is sourced in the environment, we can recreate the symmetry
 # operations table
 if(EXISTS "$ENV{CCP4}/lib/data/syminfo.lib")
-	option(CIFPP_RECREATE_SYMOP_DATA
-		"Recreate SymOp data table in case it is out of date" ON)
+	set(CIFPP_RECREATE_SYMOP_DATA ON CACHE BOOL "Recreate SymOp data table in case it is out of date")
 endif()
 
 # CCP4 build
 if(BUILD_FOR_CCP4)
 	if("$ENV{CCP4}" STREQUAL "" OR NOT EXISTS $ENV{CCP4})
-		message(FATAL_ERROR "A CCP4 built was requested but CCP4 was not sourced")
+		message(FATAL_ERROR "cifpp: A CCP4 built was requested but CCP4 was not sourced")
 	else()
 		list(PREPEND CMAKE_MODULE_PATH "$ENV{CCP4}")
 		list(PREPEND CMAKE_PREFIX_PATH "$ENV{CCP4}")
@@ -128,9 +133,6 @@ if(WIN32)
 		add_definitions(-D _WIN32_WINNT=0x0501)
 	endif()
 
-	# Man, this is 2024 we're living in...
-	add_definitions(-DNOMINMAX)
-
 	# We do not want to write an export file for all our symbols...
 	set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
 endif()
@@ -164,7 +166,7 @@ int main(int argc, char *argv[]) { return 0; }"
 
 if(GXX_LIBSTDCPP)
 	message(
-		STATUS "Testing for known regex bug, since you're using GNU libstdc++")
+		STATUS "cifpp: Testing for known regex bug, since you're using GNU libstdc++")
 
 	try_run(STD_REGEX_RUNNING STD_REGEX_COMPILING
 		${CMAKE_CURRENT_BINARY_DIR}/test
@@ -173,7 +175,7 @@ if(GXX_LIBSTDCPP)
 	if(STD_REGEX_RUNNING STREQUAL FAILED_TO_RUN)
 		message(
 			STATUS
-			"You are probably trying to compile using the g++ standard library which contains a crashing std::regex implementation. Will use boost::regex instead"
+			"cifpp: You are probably trying to compile using the g++ standard library which contains a crashing std::regex implementation. Will use boost::regex instead"
 		)
 
 		find_package(Boost 1.80 QUIET COMPONENTS regex)
@@ -221,11 +223,15 @@ if(MSVC)
 	endforeach()
 endif()
 
+if(USE_CURL_FOR_CCD)
+	find_package(CURL REQUIRED)
+endif()
+
 find_package(ZLIB QUIET)
 find_package(Threads)
 
 if(NOT ZLIB_FOUND)
-	message(FATAL_ERROR "The zlib development files were not found you this system, please install them and try again (hint: on debian/ubuntu use apt-get install zlib1g-dev)")
+	message(FATAL_ERROR "cifpp: The zlib development files were not found you this system, please install them and try again (hint: on debian/ubuntu use apt-get install zlib1g-dev)")
 endif()
 
 # Using Eigen3 is a bit of a thing. We don't want to build it completely since
@@ -242,12 +248,12 @@ else()
 		GIT_REPOSITORY https://gitlab.com/libeigen/eigen.git
 		GIT_TAG 3.4.0
 		INSTALL_COMMAND "")
-	
+
 	ExternalProject_Get_Property(my-eigen3 SOURCE_DIR)
 	set(EIGEN_INCLUDE_DIR ${SOURCE_DIR})
 endif()
 
-message(STATUS "Eigen include dir is ${EIGEN_INCLUDE_DIR}")
+message(STATUS "cifpp: Eigen include dir is ${EIGEN_INCLUDE_DIR}")
 
 # Create a revision file, containing the current git version info
 include(VersionString)
@@ -376,7 +382,13 @@ target_include_directories(
 	"$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>"
 	PRIVATE "${BOOST_REGEX_INCLUDE_DIR}" "${EIGEN_INCLUDE_DIR}")
 
-target_link_libraries(cifpp PUBLIC Threads::Threads ZLIB::ZLIB $<$<TARGET_EXISTS:std::atomic>:std::atomic>)
+target_link_libraries(cifpp
+	PUBLIC Threads::Threads ZLIB::ZLIB $<$<TARGET_EXISTS:std::atomic>:std::atomic>
+	$<IF:$<BOOL:${USE_CURL_FOR_CCD}>,CURL::libcurl,>)
+
+if (USE_CURL_FOR_CCD)
+	target_compile_definitions(cifpp PUBLIC HAVE_CURL)
+endif()
 
 if(CMAKE_CXX_COMPILER_ID STREQUAL "AppleClang")
 	target_link_options(cifpp PRIVATE -undefined dynamic_lookup)
@@ -390,7 +402,7 @@ if(CIFPP_DOWNLOAD_CCD)
 		file(SIZE ${COMPONENTS_CIF} CCD_FILE_SIZE)
 
 		if(CCD_FILE_SIZE EQUAL 0)
-			message(STATUS "Removing empty ${COMPONENTS_CIF} file")
+			message(STATUS "cifpp: Removing empty ${COMPONENTS_CIF} file")
 			file(REMOVE "${COMPONENTS_CIF}")
 		endif()
 	endif()
@@ -429,7 +441,7 @@ if(CIFPP_DOWNLOAD_CCD)
 
 		if(CCD_FETCH_STATUS_CODE)
 			message(
-				FATAL_ERROR "Error trying to download CCD file: ${CCD_FETCH_STATUS}")
+				FATAL_ERROR "cifpp: Error trying to download CCD file: ${CCD_FETCH_STATUS}")
 		endif()
 	endif()
 endif()
@@ -493,7 +505,7 @@ file(GLOB OLD_CONFIG_FILES
 
 if(OLD_CONFIG_FILES)
 	message(
-		STATUS "Installation will remove old config files: ${OLD_CONFIG_FILES}")
+		STATUS "cifpp: Installation will remove old config files: ${OLD_CONFIG_FILES}")
 	install(CODE "file(REMOVE ${OLD_CONFIG_FILES})")
 endif()
 
@@ -559,7 +571,7 @@ if(CIFPP_INSTALL_UPDATE_SCRIPT)
 			PERMISSIONS OWNER_EXECUTE OWNER_READ GROUP_EXECUTE GROUP_READ WORLD_EXECUTE
 			WORLD_READ)
 	else()
-		message(FATAL_ERROR "Don't know where to install the update script")
+		message(FATAL_ERROR "cifpp: Don't know where to install the update script")
 	endif()
 
 	# a config file, to make it complete
@@ -573,7 +585,7 @@ if(CIFPP_INSTALL_UPDATE_SCRIPT)
 		install(FILES ${CMAKE_CURRENT_BINARY_DIR}/libcifpp.conf
 			DESTINATION ${CMAKE_INSTALL_SYSCONFDIR})
 		install(
-			CODE "message(\"A configuration file has been written to ${CIFPP_ETC_DIR}/libcifpp.conf, please edit this file to enable automatic updates\")"
+			CODE "message(\"cifpp: A configuration file has been written to ${CIFPP_ETC_DIR}/libcifpp.conf, please edit this file to enable automatic updates\")"
 		)
 
 		install(DIRECTORY DESTINATION ${CMAKE_INSTALL_SYSCONFDIR}/libcifpp/cache-update.d)

changelog

@@ -1,3 +1,19 @@
+Version 9.0.0
+- Rename fields of cif::mm::polymer to match the naming
+  in mmcif_pdbx.dic. Also, related, fix building mm::structure
+  using the correct mapping between atom_site and residues.
+- _atom_site.auth_alt_id does not exist, it should be
+  _atom_site.pdbx_auth_alt_id of course.
+- Added a more lightweight fixup for mmcif_pdbx files
+  that lack certain categories.
+
+Version 8.0.1
+- Fix cif::mm::structure::cleanup_empty_categories, removed too much
+- Add default value for B_iso_or_equiv in residue::create_new_atom
+- Reconstruct some branch records in bare pdbx files
+- Fix parsing PDB files (bug due to missing validator in dest. cat.)
+- Do not fail conversion of PDB files when compound info is missing
+
 Version 8.0.0
 - A dictionary is for a datablock and a file can have
   datablocks with differing dictionaries.

cmake/cifpp-config.cmake.in

@@ -8,5 +8,6 @@ include(CMakeFindDependencyMacro)
 
 find_dependency(Threads)
 find_dependency(ZLIB REQUIRED)
+find_dependency(CURL REQUIRED)
 
 check_required_components(cifpp)

docs/basics.rst

@@ -217,7 +217,7 @@ A simple case:
     #include <cif++.hpp>
 
     cif::file f("1cbs.cif.gz");
-    f.load_dictionary("mmcif_pdbx");
+    f.load_dictionary("mmcif_pdbx.dic");
 
     if (not f.is_valid())
         std::cout << "This file is not valid\n";

include/cif++/category.hpp

@@ -32,7 +32,6 @@
 #include "cif++/iterator.hpp"
 #include "cif++/row.hpp"
 #include "cif++/text.hpp"
-#include "cif++/validate.hpp"
 
 #include <array>
 
@@ -51,6 +50,11 @@
 namespace cif
 {
 
+class validator;
+struct category_validator;
+struct item_validator;
+struct link_validator;
+
 // --------------------------------------------------------------------
 // special exceptions
 
@@ -143,6 +147,16 @@ class category
 
 	category() = default;            ///< Default constructor
 	category(std::string_view name); ///< Constructor taking a \a name
+
+	/// @brief Constructor creating a category named @a name and filled with @a rows
+	/// @param name Name for the new category
+	/// @param rows The data stored in the category
+	category(std::string_view name, row_initializer &&rows)
+		: category(name)
+	{
+		emplace(std::forward<row_initializer>(rows));
+	}
+
 	category(const category &rhs);   ///< Copy constructor
 
 	category(category &&rhs) noexcept ///< Move constructor
@@ -209,6 +223,11 @@ class category
 	/// @return Returns true is all validations pass
 	bool validate_links() const;
 
+	/**
+	 * @brief Strip removes items from this category that are invalid according to the assigned validator
+	 */
+	void strip();
+
 	/// @brief Equality operator, returns true if @a rhs is equal to this
 	/// @param rhs The object to compare with
 	/// @return True if the data contained is equal
@@ -947,6 +966,12 @@ class category
 		return insert_impl(cend(), r);
 	}
 
+	void emplace(const_iterator b, const_iterator e)
+	{
+		while (b != e)
+			emplace(*b++);
+	}
+
 	/// @brief Completely erase all rows contained in this category
 	void clear();
 
@@ -1079,25 +1104,7 @@ class category
 	// --------------------------------------------------------------------
 	/// \brief Return the index number for \a item_name
 
-	uint16_t get_item_ix(std::string_view item_name) const
-	{
-		uint16_t result;
-
-		for (result = 0; result < m_items.size(); ++result)
-		{
-			if (iequals(item_name, m_items[result].m_name))
-				break;
-		}
-
-		if (VERBOSE > 0 and result == m_items.size() and m_cat_validator != nullptr) // validate the name, if it is known at all (since it was not found)
-		{
-			auto iv = m_cat_validator->get_validator_for_item(item_name);
-			if (iv == nullptr)
-				std::cerr << "Invalid name used '" << item_name << "' is not a known item in " + m_name << '\n';
-		}
-
-		return result;
-	}
+	uint16_t get_item_ix(std::string_view item_name) const;
 
 	/// @brief Return the name for item with index @a ix
 	/// @param ix The index number
@@ -1113,28 +1120,7 @@ class category
 	/// @brief Make sure a item with name @a item_name is known and return its index number
 	/// @param item_name The name of the item
 	/// @return The index number of the item
-	uint16_t add_item(std::string_view item_name)
-	{
-		using namespace std::literals;
-
-		uint16_t result = get_item_ix(item_name);
-
-		if (result == m_items.size())
-		{
-			const item_validator *item_validator = nullptr;
-
-			if (m_cat_validator != nullptr)
-			{
-				item_validator = m_cat_validator->get_validator_for_item(item_name);
-				if (item_validator == nullptr)
-					m_validator->report_error(validation_error::item_not_allowed_in_category, m_name, item_name, false);
-			}
-
-			m_items.emplace_back(item_name, item_validator);
-		}
-
-		return result;
-	}
+	uint16_t add_item(std::string_view item_name);
 
 	/** @brief Remove item name @a colum_name
 	 * @param item_name The item to be removed

include/cif++/compound.hpp

@@ -196,6 +196,23 @@ class compound
 // --------------------------------------------------------------------
 // Factory class for compound and Link objects
 
+/// @brief Options available to configure a compound factory
+struct compound_factory_options
+{
+	/// If you have a multithreaded application and want to have different
+	/// compounds in each thread (e.g. a web service processing user requests
+	/// with different sets of compounds) you can set this flag to true.
+	bool use_thread_local_instance_only = false;
+
+#if HAVE_CURL
+	// Various locations for chem_comp data files:
+	// - ftp://files.ebi.ac.uk/pub/databases/pdb/refdata/chem_comp
+	// - https://files.rcsb.org/pub/pdb/refdata/chem_comp/
+
+	std::string remote_chem_comp_url = "ftp://files.ebi.ac.uk/pub/databases/pdb/refdata/chem_comp";
+#endif
+};
+
 /// Use the compound_factory singleton instance to create compound objects
 
 class compound_factory
@@ -208,8 +225,12 @@ class compound_factory
 	/// with different sets of compounds) you can set the \a useThreadLocalInstanceOnly
 	/// flag to true.
 
+	[[deprecated("Use version with compound_factory_options instead")]]
 	static void init(bool useThreadLocalInstanceOnly);
 
+	/// \brief Initialise a singleton instance.
+	static void init(compound_factory_options options = {});
+
 	/// Return the singleton instance. If initialized with local threads, this is the
 	/// instance for the current thread.
 	static compound_factory &instance();
@@ -239,6 +260,8 @@ class compound_factory
 	void push_dictionary(const file &file);
 
 	/// Remove the last pushed dictionary
+
+	// TODO: check if the popped dict is the correct one
 	void pop_dictionary();
 
 	/// Return whether @a res_name is a valid and known peptide
@@ -298,7 +321,7 @@ class compound_factory
 
 	static std::unique_ptr<compound_factory> s_instance;
 	static thread_local std::unique_ptr<compound_factory> tl_instance;
-	static bool s_use_thread_local_instance;
+	static compound_factory_options s_options;
 
 	std::shared_ptr<compound_factory_impl> m_impl;
 };
@@ -320,6 +343,9 @@ class compound_factory
  * @endcode
  */
 
+
+// TODO: check if pushed and popped dicts are the same!
+
 class compound_source
 {
   public:

include/cif++/datablock.hpp

@@ -29,6 +29,8 @@
 #include "cif++/category.hpp"
 #include "cif++/forward_decl.hpp"
 
+#include <list>
+
 /** \file datablock.hpp
  * Each valid mmCIF file contains at least one @ref cif::datablock.
  * A datablock has a name and can contain one or more @ref cif::category "categories"
@@ -104,13 +106,6 @@ class datablock : public std::list<category>
 	 */
 	void load_dictionary();
 
-	/**
-	 * @brief Load the dictionary named @a dict_name
-	 * 
-	 * @param dict_name 
-	 */
-	void load_dictionary(std::string_view dict_name);
-
 	/**
 	 * @brief Set the validator object to @a v
 	 * 
@@ -133,15 +128,6 @@ class datablock : public std::list<category>
 	 */
 	bool is_valid() const;
 
-	/**
-	 * @brief Validates the content of this datablock and all its content
-	 * and updates or removes the audit_conform category to match the result.
-	 * 
-	 * @return true If the content is valid
-	 * @return false If the content is not valid
-	 */
-	bool is_valid();
-
 	/**
 	 * @brief Validates all contained data for valid links between parents and children
 	 * as defined in the validator
@@ -151,6 +137,14 @@ class datablock : public std::list<category>
 	 */
 	bool validate_links() const;
 
+	/**
+	 * @brief Strip removes all categories and items that are invalid according
+	 * to the assigned validator. Will also add a valid audit_conform block.
+	 * 
+	 * @return true if the remaining datablock is valid
+	 */
+	bool strip();
+
 	// --------------------------------------------------------------------
 
 	/**
@@ -189,6 +183,15 @@ class datablock : public std::list<category>
 	 */
 	const category *get(std::string_view name) const;
 
+
+	/**
+	 * @brief Return true if this datablock contains a non-empty category
+	 */
+	bool contains(std::string_view name) const
+	{
+		return get(name) != nullptr;
+	}
+
 	/**
 	 * @brief Tries to find a category with name @a name and will create a
 	 * new one if it is not found. The result is a tuple of an iterator
@@ -247,4 +250,4 @@ class datablock : public std::list<category>
 	const validator *m_validator = nullptr;
 };
 
-} // namespace cif
+} // namespace cif

include/cif++/dictionary_parser.hpp

@@ -38,13 +38,8 @@ namespace cif
 {
 
 /**
- * @brief Parse the contents of @a is and create a new validator object with name @a name
+ * @brief Parse the contents of @a is and place content in validator @a v
  */
-validator parse_dictionary(std::string_view name, std::istream &is);
-
-/**
- * @brief Extend the definitions in validator @a v with the contents of stream @a is
- */
-void extend_dictionary(validator &v, std::istream &is);
+void parse_dictionary(validator &v, std::istream &is);
 
 } // namespace cif

include/cif++/file.hpp

@@ -210,12 +210,6 @@ class file : public std::list<datablock>
 	/** Load the data from @a is using validator @a v */
 	void load(std::istream &is, const validator &v);
 
-	/** Load the data from the file specified by @a p using a validator constructed from dictionary @a dict */
-	void load(const std::filesystem::path &p, std::string_view dict);
-
-	/** Load the data from @a is using a validator constructed from dictionary @a dict */
-	void load(std::istream &is, std::string_view dict);
-
 	/** Save the data to the file specified by @a p */
 	void save(const std::filesystem::path &p) const;
 
@@ -232,4 +226,4 @@ class file : public std::list<datablock>
 	}
 };
 
-} // namespace cif
+} // namespace cif

include/cif++/model.hpp

@@ -34,7 +34,7 @@
 #include <numeric>
 
 #if __cpp_lib_format
-#include <format>
+# include <format>
 #endif
 
 /** @file model.hpp
@@ -106,8 +106,6 @@ class atom
 
 		atom_impl(const atom_impl &i) = default;
 
-		void prefetch();
-
 		int compare(const atom_impl &b) const;
 
 		// bool getAnisoU(float anisou[6]) const;
@@ -345,7 +343,7 @@ class atom
 	std::string get_auth_asym_id() const { return get_property("auth_asym_id"); }      ///< Return the auth_asym_id property
 	std::string get_auth_seq_id() const { return get_property("auth_seq_id"); }        ///< Return the auth_seq_id property
 	std::string get_auth_atom_id() const { return get_property("auth_atom_id"); }      ///< Return the auth_atom_id property
-	std::string get_auth_alt_id() const { return get_property("auth_alt_id"); }        ///< Return the auth_alt_id property
+	std::string get_auth_alt_id() const { return get_property("pdbx_auth_alt_id"); }   ///< Return the auth_alt_id property
 	std::string get_auth_comp_id() const { return get_property("auth_comp_id"); }      ///< Return the auth_comp_id property
 	std::string get_pdb_ins_code() const { return get_property("pdbx_PDB_ins_code"); } ///< Return the pdb_ins_code property
 
@@ -481,8 +479,8 @@ class residue
 		, m_compound_id(compoundID)
 		, m_asym_id(asymID)
 		, m_seq_id(seqID)
-		, m_auth_asym_id(authAsymID)
-		, m_auth_seq_id(authSeqID)
+		, m_pdb_strand_id(authAsymID)
+		, m_pdb_seq_num(authSeqID)
 		, m_pdb_ins_code(pdbInsCode)
 	{
 	}
@@ -509,9 +507,9 @@ class residue
 	const std::string &get_asym_id() const { return m_asym_id; } ///< Return the asym_id
 	int get_seq_id() const { return m_seq_id; }                  ///< Return the seq_id
 
-	const std::string get_auth_asym_id() const { return m_auth_asym_id; } ///< Return the auth_asym_id
-	const std::string get_auth_seq_id() const { return m_auth_seq_id; }   ///< Return the auth_seq_id
-	std::string get_pdb_ins_code() const { return m_pdb_ins_code; }       ///< Return the pdb_ins_code
+	const std::string get_pdb_strand_id() const { return m_pdb_strand_id; } ///< Return the pdb_strand_id
+	const std::string get_pdb_seq_num() const { return m_pdb_seq_num; }     ///< Return the pdb_seq_num
+	std::string get_pdb_ins_code() const { return m_pdb_ins_code; }         ///< Return the pdb_ins_code
 
 	const std::string &get_compound_id() const { return m_compound_id; } ///< Return the compound_id
 	void set_compound_id(const std::string &id) { m_compound_id = id; }  ///< Set the compound_id to @a id
@@ -540,6 +538,9 @@ class residue
 	/// \brief Return the atom with atom_id @a atomID
 	atom get_atom_by_atom_id(const std::string &atomID) const;
 
+	/// \brief Return the atom with atom_id @a atomID and alternate_id @a altID
+	atom get_atom_by_atom_id(const std::string &atomID, const std::string &altID) const;
+
 	/// \brief Return the list of atoms having ID \a atomID
 	///
 	/// This includes all alternate atoms with this ID
@@ -577,7 +578,7 @@ class residue
 								   m_seq_id == rhs.m_seq_id and
 								   m_asym_id == rhs.m_asym_id and
 								   m_compound_id == rhs.m_compound_id and
-								   m_auth_seq_id == rhs.m_auth_seq_id);
+								   m_pdb_seq_num == rhs.m_pdb_seq_num);
 	}
 
 	/// @brief Create a new atom and add it to the list
@@ -591,7 +592,7 @@ class residue
 	structure *m_structure = nullptr;
 	std::string m_compound_id, m_asym_id;
 	int m_seq_id = 0;
-	std::string m_auth_asym_id, m_auth_seq_id, m_pdb_ins_code;
+	std::string m_pdb_strand_id, m_pdb_seq_num, m_pdb_ins_code;
 	std::vector<atom> m_atoms;
 	/** @endcond */
 };
@@ -622,6 +623,9 @@ class monomer : public residue
 	bool is_first_in_chain() const; ///< Return if this residue is the first residue in the chain
 	bool is_last_in_chain() const;  ///< Return if this residue is the last residue in the chain
 
+	const monomer &prev() const; // Return previous monomer in polymer
+	const monomer &next() const; // Return next monomer in polymer
+
 	// convenience
 	bool has_alpha() const; ///< Return if a alpha value can be calculated (depends on location in chain)
 	bool has_kappa() const; ///< Return if a kappa value can be calculated (depends on location in chain)
@@ -708,15 +712,15 @@ class polymer : public std::vector<monomer>
 
 	structure *get_structure() const { return m_structure; } ///< Return the structure
 
-	std::string get_asym_id() const { return m_asym_id; }           ///< Return the asym_id
-	std::string get_auth_asym_id() const { return m_auth_asym_id; } ///< Return the PDB chain ID, actually
-	std::string get_entity_id() const { return m_entity_id; }       ///< Return the entity_id
+	std::string get_asym_id() const { return m_asym_id; }             ///< Return the asym_id
+	std::string get_pdb_strand_id() const { return m_pdb_strand_id; } ///< Return the PDB chain ID, actually
+	std::string get_entity_id() const { return m_entity_id; }         ///< Return the entity_id
 
   private:
 	structure *m_structure;
 	std::string m_entity_id;
 	std::string m_asym_id;
-	std::string m_auth_asym_id;
+	std::string m_pdb_strand_id;
 };
 
 // --------------------------------------------------------------------
@@ -754,7 +758,7 @@ class sugar : public residue
 	int num() const
 	{
 		int result;
-		auto r = std::from_chars(m_auth_seq_id.data(), m_auth_seq_id.data() + m_auth_seq_id.length(), result);
+		auto r = std::from_chars(m_pdb_seq_num.data(), m_pdb_seq_num.data() + m_pdb_seq_num.length(), result);
 		if ((bool)r.ec)
 			throw std::runtime_error("The auth_seq_id should be a number for a sugar");
 		return result;
@@ -853,19 +857,38 @@ class branch : public std::vector<sugar>
 	std::string m_asym_id, m_entity_id;
 };
 
-// --------------------------------------------------------------------
-
-/// \brief A still very limited set of options for reading structures
-enum class StructureOpenOptions
+/** @brief Enumeration for controlling atom selection based on occupancy. */
+enum class occupancy_policy
 {
-	SkipHydrogen = 1 << 0 ///< Do not include hydrogen atoms in the structure object
+	/** @brief Include all atoms regardless of their occupancy factor. */
+	ALL = 0,
+
+	/** @brief Select only alternate atoms with the maximum occupancy factor.
+	 * If multiple atoms have the same maximum occupancy, choose the one with the minimum B-factor.
+	 * If multiple atoms share both the maximum occupancy and the minimum B-factor, select the first encountered atom.
+	 */
+	MAX = 1,
+
+	/** @brief Select only alternate atoms with the minimum occupancy factor.
+	 * Similar to MAX, if multiple atoms have the same minimum occupancy, choose the one with the minimum B-factor.
+	 * If multiple atoms share both the minimum occupancy and the minimum B-factor, select the first encountered atom.
+	 */
+	MIN = 2,
+
+	/** @brief Exclude all atoms with an occupancy factor greater than zero. */
+	UNOCCUPIED = 3
 };
 
-/// \brief A way to combine two options. Not very useful as there is only one...
-constexpr inline bool operator&(StructureOpenOptions a, StructureOpenOptions b)
+struct structure_open_options
 {
-	return static_cast<int>(a) bitand static_cast<int>(b);
-}
+	bool skip_hydrogen = false;                              ///< Do not include hydrogen atoms in the structure object
+	bool skip_hetatom = false;                               ///< Do not include HET atoms in the structure object
+	bool skip_water = false;                                 ///< Do not include water atoms in the structure object
+	occupancy_policy occupancy_mode = occupancy_policy::ALL; ///< By default, the occupancy policy is set to occupancy_policy::ALL
+	std::vector<std::string> asyms;                          ///< The asyms to load, if empty load all
+	std::optional<float> min_b_factor;                       ///< Only load atoms with at least this b_factor
+	std::optional<float> max_b_factor;                       ///< Only load atoms with at most this b_factor
+};
 
 // --------------------------------------------------------------------
 
@@ -879,10 +902,10 @@ class structure
 {
   public:
 	/// \brief Read the structure from cif::file @a p
-	structure(file &p, std::size_t modelNr = 1, StructureOpenOptions options = {});
+	structure(file &p, std::size_t modelNr = 1, structure_open_options options = {});
 
 	/// \brief Load the structure from already parsed mmCIF data in @a db
-	structure(datablock &db, std::size_t modelNr = 1, StructureOpenOptions options = {});
+	structure(datablock &db, std::size_t modelNr = 1, structure_open_options options = {});
 
 	/** @cond */
 	structure(structure &&s) = default;
@@ -992,18 +1015,18 @@ class structure
 	/**
 	 * @brief Change residue @a res to a new compound ID optionally
 	 * remapping atoms.
-	 * 
+	 *
 	 * A new chem_comp entry as well as an entity is created if needed and
 	 * if the list of @a remappedAtoms is not empty it is used to remap.
-	 * 
+	 *
 	 * The array in @a remappedAtoms contains tuples of strings, both
 	 * strings contain an atom_id. The first is the one in the current
 	 * residue and the second is the atom_id that should be used instead.
 	 * If the second string is empty, the atom is removed from the residue.
-	 * 
-	 * @param res 
-	 * @param newcompound 
-	 * @param remappedAtoms 
+	 *
+	 * @param res
+	 * @param newcompound
+	 * @param remappedAtoms
 	 */
 	void change_residue(residue &res, const std::string &newcompound,
 		const std::vector<std::tuple<std::string, std::string>> &remappedAtoms);
@@ -1112,7 +1135,7 @@ class structure
 	friend polymer;
 	friend residue;
 
-	void load_atoms_for_model(StructureOpenOptions options);
+	void load_atoms_for_model(structure_open_options options);
 
 	std::string insert_compound(const std::string &compoundID, bool is_entity);
 

include/cif++/pdb.hpp

@@ -27,6 +27,7 @@
 #pragma once
 
 #include "cif++/file.hpp"
+#include "cif++/validate.hpp"
 
 #include <system_error>
 
@@ -103,16 +104,50 @@ inline void write(const std::filesystem::path &p, const file &f)
 
 // --------------------------------------------------------------------
 
+/**
+ * @brief Quickly fix a PDB file that lacks some often needed categories
+ * 
+ * This differs from reconstruct_pdbx which does a much more thorough job
+ * 
+ * \param pdbx_file The cif::file that hopefully contains some valid data
+ */
+
+void fixup_pdbx(file &pdbx_file);
+
+/**
+ * @brief Quickly fix a PDB file that lacks some often needed categories
+ * 
+ * This differs from reconstruct_pdbx which does a much more thorough job
+ * 
+ * \param pdbx_file The cif::file that hopefully contains some valid data
+ * \param v The validator to use
+ */
+
+void fixup_pdbx(file &pdbx_file, const validator &v);
+
 /** \brief Reconstruct all missing categories for an assumed PDBx file.
  *
  * Some people believe that simply dumping some atom records is enough.
+ * 
+ * This version uses the audit_conform information and falls back to
+ * using mmcif_pdbx.dic if not specified.
  *
- * \param file The cif::file that hopefully contains some valid data
- * \param dictionary The mmcif dictionary to use
+ * \param pdbx_file The cif::file that hopefully contains some valid data
  * \result Returns true if the resulting file is valid
  */
 
-bool reconstruct_pdbx(file &pdbx_file, std::string_view dictionary = "mmcif_pdbx");
+ bool reconstruct_pdbx(file &pdbx_file);
+
+ /** \brief Reconstruct all missing categories for an assumed PDBx file.
+ *
+ * Some people believe that simply dumping some atom records is enough.
+ *
+ * \param pdbx_file The cif::file that hopefully contains some valid data
+ * \param v The validator to use
+ * \result Returns true if the resulting file is valid
+ */
+
+bool reconstruct_pdbx(file &pdbx_file, const validator &v);
 
 /** \brief This is an extension to cif::validator, use the logic in common
  * PDBx files to see if the file is internally consistent.
@@ -125,12 +160,13 @@ bool reconstruct_pdbx(file &pdbx_file, std::string_view dictionary = "mmcif_pdbx
  * 
  * This function throws a std::system_error in case of an error
  *
- * \param file The input file
- * \param dictionary The mmcif dictionary to use
+ * \param pdbx_file The input file
+ * \param v The validator to use
  * \result Returns true if the file was valid and consistent
  */
 
-bool is_valid_pdbx_file(const file &pdbx_file, std::string_view dictionary = "mmcif_pdbx");
+bool is_valid_pdbx_file(const file &pdbx_file,
+	const validator &v = validator_factory::instance().get("mmcif_pdbx.dic"));
 
 /** \brief This is an extension to cif::validator, use the logic in common
  * PDBx files to see if the file is internally consistent.
@@ -145,7 +181,6 @@ bool is_valid_pdbx_file(const file &pdbx_file, std::string_view dictionary = "mm
  * default mmcif_pdbx.dic dictionary.
  *
  * \param file The input file
- * \param ec The error_code in case something was wrong
  * \result Returns true if the file was valid and consistent
  */
 
@@ -161,12 +196,12 @@ bool is_valid_pdbx_file(const file &pdbx_file, std::error_code &ec);
  * Use the common \ref cif::VERBOSE flag to turn on diagnostic messages.
  * 
  * \param file The input file
- * \param dictionary The dictionary to use
+ * \param v The validator to use
  * \param ec The error_code in case something was wrong
  * \result Returns true if the file was valid and consistent
  */
 
-bool is_valid_pdbx_file(const file &pdbx_file, std::string_view dictionary,
+bool is_valid_pdbx_file(const file &pdbx_file, const validator &v,
 	std::error_code &ec);
 
 // --------------------------------------------------------------------

include/cif++/validate.hpp

@@ -26,12 +26,14 @@
 
 #pragma once
 
+#include "cif++/category.hpp"
 #include "cif++/text.hpp"
 
 #include <cassert>
 #include <filesystem>
 #include <list>
 #include <mutex>
+#include <optional>
 #include <system_error>
 #include <utility>
 
@@ -48,6 +50,7 @@
 namespace cif
 {
 
+class category;
 struct category_validator;
 
 // --------------------------------------------------------------------
@@ -227,38 +230,39 @@ struct type_validator
 {
 	std::string m_name;                 ///< The name of the type
 	DDL_PrimitiveType m_primitive_type; ///< The primitive_type of the type
-	regex_impl *m_rx;                   ///< The regular expression for the type
+	std::shared_ptr<regex_impl> m_rx;   ///< The regular expression for the type
 
 	type_validator() = delete;
 
 	/// @brief Constructor
 	type_validator(std::string_view name, DDL_PrimitiveType type, std::string_view rx);
 
-	type_validator(const type_validator &) = delete;
-
 	/// @brief Copy constructor
-	type_validator(type_validator &&rhs)
-		: m_name(std::move(rhs.m_name))
-		, m_primitive_type(rhs.m_primitive_type)
-	{
-		m_rx = std::exchange(rhs.m_rx, nullptr);
-	}
-
-	type_validator &operator=(const type_validator &) = delete;
+	type_validator(const type_validator &tv);
 
 	/// @brief Move constructor
-	type_validator &operator=(type_validator &&rhs)
+	type_validator(type_validator &&rhs)
 	{
-		m_name = std::move(rhs.m_name);
-		m_primitive_type = rhs.m_primitive_type;
-		m_rx = std::exchange(rhs.m_rx, nullptr);
+		swap(*this, rhs);
+	}
 
+	/// @brief Move constructor
+	type_validator &operator=(type_validator rhs)
+	{
+		swap(*this, rhs);
 		return *this;
 	}
 
 	/// @brief Destructor
 	~type_validator();
 
+	friend void swap(type_validator &a, type_validator &b)
+	{
+		std::swap(a.m_name, b.m_name);
+		std::swap(a.m_primitive_type, b.m_primitive_type);
+		std::swap(a.m_rx, b.m_rx);
+	}
+
 	/// @brief Return the sorting order
 	bool operator<(const type_validator &rhs) const
 	{
@@ -303,13 +307,13 @@ struct item_alias
  */
 struct item_validator
 {
-	std::string m_item_name;                  ///< The item name
-	bool m_mandatory;                         ///< Flag indicating this item is mandatory
-	const type_validator *m_type;             ///< The type for this item
-	cif::iset m_enums;                        ///< If filled, the set of allowed values
-	std::string m_default;                    ///< If filled, a default value for this item
-	category_validator *m_category = nullptr; ///< The category_validator this item_validator belongs to
-	std::vector<item_alias> m_aliases;        ///< The aliases for this item
+	std::string m_item_name;           ///< The item name
+	bool m_mandatory;                  ///< Flag indicating this item is mandatory
+	const type_validator *m_type;      ///< The type for this item
+	cif::iset m_enums;                 ///< If filled, the set of allowed values
+	std::string m_default;             ///< If filled, a default value for this item
+	std::string m_category;            ///< The category this item_validator belongs to
+	std::vector<item_alias> m_aliases; ///< The aliases for this item
 
 	/// @brief Compare based on the name
 	bool operator<(const item_validator &rhs) const
@@ -396,24 +400,48 @@ class validator
 	 *
 	 * @param name The name of the underlying dictionary
 	 */
-	validator(std::string_view name)
-		: m_name(name)
+	validator()
+		: m_audit_conform("audit_conform")
 	{
 	}
 
+	/**
+	 * @brief Construct a new validator object
+	 *
+	 * @param name The name of the underlying dictionary
+	 * @param is The data to parse
+	 */
+	validator(std::istream &is)
+		: m_audit_conform("audit_conform")
+	{
+		parse(is);
+	}
+
 	/// @brief destructor
 	~validator() = default;
 
-	validator(const validator &rhs) = delete;
-	validator &operator=(const validator &rhs) = delete;
+	validator(const validator &rhs);
 
 	/// @brief move constructor
-	validator(validator &&rhs) = default;
+	validator(validator &&rhs)
+	{
+		swap(*this, rhs);
+	}
 
-	/// @brief move assignment operator
-	validator &operator=(validator &&rhs) = default;
+	validator &operator=(validator rhs)
+	{
+		swap(*this, rhs);
+		return *this;
+	}
+
+	friend void swap(validator &a, validator &b) noexcept;
 
 	friend class dictionary_parser;
+	friend class validator_factory;
+
+	/// @brief Parse dictionary in @a is and put content in this validator, optionally extending it
+	/// @param is The stream containing a valid cif dictionary
+	void parse(std::istream &is);
 
 	/// @brief Add type_validator @a v to the list of type validators
 	void add_type_validator(type_validator &&v);
@@ -456,18 +484,22 @@ class validator
 	void report_error(std::error_code ec, std::string_view category,
 		std::string_view item, bool fatal = true) const;
 
-	const std::string &name() const { return m_name; }        ///< Get the name of this validator
-	void set_name(const std::string &name) { m_name = name; } ///< Set the name of this validator
+	/// @brief Write out the audit_conform data for this validator
+	/// @param audit_conform
+	void fill_audit_conform(category &audit_conform) const;
 
-	const std::string &version() const { return m_version; }              ///< Get the version of this validator
-	void set_version(const std::string &version) { m_version = version; } ///< Set the version of this validator
+	/// @brief Return true if this validator matches @a audit_conform
+	bool matches_audit_conform(const category &audit_conform) const;
+
+	/// @brief Add info
+	void append_audit_conform(const std::string &name, const std::optional<std::string> &version);
 
   private:
 	// name is fully qualified here:
 	item_validator *get_validator_for_item(std::string_view name) const;
 
-	std::string m_name;
-	std::string m_version;
+	category m_audit_conform;
+
 	bool m_strict = false;
 	std::set<type_validator> m_type_validators;
 	std::set<category_validator> m_category_validators;
@@ -487,17 +519,28 @@ class validator_factory
 	/// @brief Return the singleton instance
 	static validator_factory &instance();
 
-	/// @brief Return the validator with name @a dictionary_name
-	const validator &operator[](std::string_view dictionary_name);
+	/// @brief Return validator with info recorded in @a audit_conform
+	const validator &get(const category &audit_conform);
 
-	/// @brief Construct a new validator with name @a name from the data in @a is
-	const validator &construct_validator(std::string_view name, std::istream &is);
+	/// @brief Return the single-file validator with name @a dictionary_name
+	const validator &get(std::string_view dictionary_name);
+
+	/// @brief Return true if the version @a found is equal or higher than @a expected for dictionary @a name
+	static bool check_version(std::string_view name, std::string_view expected, std::string_view found);
+
+	/// @brief Add validator @a v to the list of known validators
+	const validator &add(validator &&v)
+	{
+		std::unique_lock lock(m_mutex);
+		return m_validators.emplace_back(std::move(v));
+	}
 
   private:
-	// --------------------------------------------------------------------
-
 	validator_factory() = default;
 
+	/// @brief Construct a new validator with name @a name from the data in @a is with at least version @a version if specified
+	validator construct_validator(std::string_view name, std::optional<std::string> version);
+
 	std::mutex m_mutex;
 	std::list<validator> m_validators;
 };

src/category.cpp

@@ -28,6 +28,7 @@
 #include "cif++/datablock.hpp"
 #include "cif++/parser.hpp"
 #include "cif++/utilities.hpp"
+#include "cif++/validate.hpp"
 
 #include <numeric>
 #include <stack>
@@ -542,6 +543,49 @@ category::~category()
 
 // --------------------------------------------------------------------
 
+uint16_t category::get_item_ix(std::string_view item_name) const
+{
+	uint16_t result;
+
+	for (result = 0; result < m_items.size(); ++result)
+	{
+		if (iequals(item_name, m_items[result].m_name))
+			break;
+	}
+
+	if (VERBOSE > 0 and result == m_items.size() and m_cat_validator != nullptr) // validate the name, if it is known at all (since it was not found)
+	{
+		auto iv = m_cat_validator->get_validator_for_item(item_name);
+		if (iv == nullptr)
+			std::cerr << "Invalid name used '" << item_name << "' is not a known item in " + m_name << '\n';
+	}
+
+	return result;
+}
+
+uint16_t category::add_item(std::string_view item_name)
+{
+	using namespace std::literals;
+
+	uint16_t result = get_item_ix(item_name);
+
+	if (result == m_items.size())
+	{
+		const item_validator *item_validator = nullptr;
+
+		if (m_cat_validator != nullptr)
+		{
+			item_validator = m_cat_validator->get_validator_for_item(item_name);
+			if (item_validator == nullptr)
+				m_validator->report_error(validation_error::item_not_allowed_in_category, m_name, item_name, false);
+		}
+
+		m_items.emplace_back(item_name, item_validator);
+	}
+
+	return result;
+}
+
 void category::remove_item(std::string_view item_name)
 {
 	for (std::size_t ix = 0; ix < m_items.size(); ++ix)
@@ -870,6 +914,24 @@ bool category::validate_links() const
 	return result;
 }
 
+void category::strip()
+{
+	std::vector<std::string> to_be_removed;
+
+	for (auto &item : m_items)
+	{
+		if (item.m_validator == nullptr)
+			to_be_removed.push_back(item.m_name);
+	}
+
+	for (auto item : to_be_removed)
+	{
+		if (cif::VERBOSE > 0)
+			std::clog << "Dropping item " << m_name << '.' << item << '\n';
+		remove_item(item);
+	}
+}
+
 // --------------------------------------------------------------------
 
 row_handle category::operator[](const key_type &key)
@@ -2173,4 +2235,4 @@ bool category::operator==(const category &rhs) const
 	return true;
 }
 
-} // namespace cif
+} // namespace cif

src/compound.cpp

@@ -26,6 +26,10 @@
 
 #include "cif++.hpp"
 
+#if HAVE_CURL
+# include <curl/curl.h>
+#endif
+
 #include <filesystem>
 #include <fstream>
 #include <map>
@@ -140,7 +144,7 @@ compound::compound(cif::datablock &db)
 
 	cif::tie(m_id, m_name, m_type, m_formula, m_formula_weight, m_formal_charge, one_letter_code, m_parent_id) =
 		chemComp.front().get("id", "name", "type", "formula", "formula_weight", "pdbx_formal_charge", "one_letter_code", "mon_nstd_parent_comp_id");
-	
+
 	if (one_letter_code.length() == 1)
 		m_one_letter_code = one_letter_code.front();
 
@@ -159,7 +163,7 @@ compound::compound(cif::datablock &db)
 		if (stereo_config.empty())
 			atom.stereo_config = stereo_config_type::N;
 		else
-		atom.stereo_config = parse_stereo_config_from_string(stereo_config);
+			atom.stereo_config = parse_stereo_config_from_string(stereo_config);
 		m_atoms.push_back(std::move(atom));
 	}
 
@@ -172,7 +176,7 @@ compound::compound(cif::datablock &db)
 		if (valueOrder.empty())
 			bond.type = bond_type::sing;
 		else
-		bond.type = parse_bond_type_from_string(valueOrder);
+			bond.type = parse_bond_type_from_string(valueOrder);
 		m_bonds.push_back(std::move(bond));
 	}
 }
@@ -231,12 +235,12 @@ float compound::bond_length(const std::string &atomId_1, const std::string &atom
 
 bool compound::is_peptide() const
 {
-	return iequals(m_type, "l-peptide linking")	or iequals(m_type, "peptide linking");
+	return iequals(m_type, "l-peptide linking") or iequals(m_type, "peptide linking");
 }
 
 bool compound::is_base() const
 {
-	return iequals(m_type, "dna linking")	or iequals(m_type, "rna linking");
+	return iequals(m_type, "dna linking") or iequals(m_type, "rna linking");
 }
 
 // --------------------------------------------------------------------
@@ -300,36 +304,13 @@ class compound_factory_impl : public std::enable_shared_from_this<compound_facto
 
 		compound *result = nullptr;
 
-		// walk the list, see if any of the implementations has the compound already
 		for (auto impl = shared_from_this(); impl; impl = impl->m_next)
 		{
-			for (auto cmp : impl->m_compounds)
-			{
-				if (iequals(cmp->id(), id))
-				{
-					result = cmp;
-					break;
-				}
-			}
-
-			if (result)
+			result = impl->create(id);
+			if (result != nullptr)
 				break;
 		}
 
-		if (result == nullptr and
-			find_if(m_missing.begin(), m_missing.end(), [&id](const std::string &m_id) { return cif::iequals(id, m_id); }) == m_missing.end())
-		{
-			for (auto impl = shared_from_this(); impl; impl = impl->m_next)
-			{
-				result = impl->create(id);
-				if (result != nullptr)
-					break;
-			}
-
-			if (result == nullptr)
-				m_missing.emplace_back(id);
-		}
-
 		return result;
 	}
 
@@ -360,7 +341,7 @@ class compound_factory_impl : public std::enable_shared_from_this<compound_facto
 	cif::parser::datablock_index m_index;
 
 	std::vector<compound *> m_compounds;
-	std::vector<std::string> m_missing;
+	cif::iset m_missing;
 	std::shared_ptr<compound_factory_impl> m_next;
 };
 
@@ -381,6 +362,16 @@ compound_factory_impl::compound_factory_impl(const fs::path &file, std::shared_p
 
 compound *compound_factory_impl::create(const std::string &id)
 {
+	// shortcut
+
+	if (m_missing.contains(id))
+		return nullptr;
+
+	if (auto i = find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c)
+			{ return c->id() == id; });
+		i != m_compounds.end())
+		return *i;
+
 	compound *result = nullptr;
 
 	std::unique_ptr<std::istream> ccd;
@@ -449,6 +440,9 @@ compound *compound_factory_impl::create(const std::string &id)
 		}
 	}
 
+	if (result == nullptr)
+		m_missing.insert(id);
+
 	return result;
 }
 
@@ -461,26 +455,11 @@ class local_compound_factory_impl : public compound_factory_impl
 		: compound_factory_impl(next)
 		, m_local_file(file)
 	{
-		// const std::regex peptideRx("(?:[lmp]-)?peptide", std::regex::icase);
-
-		// for (const auto &[id, name, threeLetterCode, group] :
-		// 	file["comp_list"]["chem_comp"].rows<std::string, std::string, std::string, std::string>("id", "name", "three_letter_code", "group"))
-		// {
-		// 	auto &rdb = m_local_file["comp_" + id];
-		// 	if (rdb.empty())
-		// 	{
-		// 		// std::cerr << "Missing data in restraint file for id " + id + '\n';
-		// 		continue;
-		// 	}
-
-		// 	construct_compound(rdb, id, name, threeLetterCode, group);
-		// }
 	}
 
 	compound *create(const std::string &id) override;
 
   private:
-
 	compound *construct_compound(const datablock &db, const std::string &id, const std::string &name, const std::string &three_letter_code, const std::string &group);
 
 	cif::file m_local_file;
@@ -488,6 +467,14 @@ class local_compound_factory_impl : public compound_factory_impl
 
 compound *local_compound_factory_impl::create(const std::string &id)
 {
+	if (m_missing.contains(id))
+		return nullptr;
+
+	if (auto i = find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c)
+			{ return c->id() == id; });
+		i != m_compounds.end())
+		return *i;
+
 	compound *result = nullptr;
 
 	for (auto &db : m_local_file)
@@ -514,6 +501,9 @@ compound *local_compound_factory_impl::create(const std::string &id)
 		}
 	}
 
+	if (result == nullptr)
+		m_missing.insert(id);
+
 	return result;
 }
 
@@ -524,27 +514,27 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 
 	float formula_weight = 0;
 	int formal_charge = 0;
-	std::map<std::string,std::size_t> formula_data;
+	std::map<std::string, std::size_t> formula_data;
 
-	for (std::size_t ord = 1; const auto &[atom_id, type_symbol, type, charge, x, y, z] :
-		rdb["chem_comp_atom"].rows<std::string, std::string, std::string, int, float, float, float>(
-			"atom_id", "type_symbol", "type", "charge", "x", "y", "z"))
+	for (std::size_t ord = 1; const auto &[atom_id, type_symbol, type, charge, x, y, z, xi, yi, zi] :
+		rdb["chem_comp_atom"].rows<std::string, std::string, std::string, int, std::optional<float>, std::optional<float>, std::optional<float>, std::optional<float>, std::optional<float>, std::optional<float>>(
+			"atom_id", "type_symbol", "type", "charge",
+			"model_Cartn_x", "model_Cartn_y", "model_Cartn_z",
+			"pdbx_model_Cartn_x_ideal", "pdbx_model_Cartn_y_ideal", "pdbx_model_Cartn_z_ideal"))
 	{
 		auto atom = cif::atom_type_traits(type_symbol);
 		formula_weight += atom.weight();
 
 		formula_data[type_symbol] += 1;
 
-		db["chem_comp_atom"].emplace({
-			{ "comp_id", id },
+		db["chem_comp_atom"].emplace({ { "comp_id", id },
 			{ "atom_id", atom_id },
 			{ "type_symbol", type_symbol },
 			{ "charge", charge },
-			{ "model_Cartn_x",  x, 3 },
-			{ "model_Cartn_y",  y, 3 },
-			{ "model_Cartn_z",  z, 3 },
-			{ "pdbx_ordinal", ord++ }
-		});
+			{ "model_Cartn_x", x.has_value() ? x : xi, 3 },
+			{ "model_Cartn_y", y.has_value() ? y : yi, 3 },
+			{ "model_Cartn_z", z.has_value() ? z : zi, 3 },
+			{ "pdbx_ordinal", ord++ } });
 
 		formal_charge += charge;
 	}
@@ -561,21 +551,19 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 		else if (cif::iequals(type, "triple") or cif::iequals(type, "trip"))
 			value_order = "TRIP";
 
-		db["chem_comp_bond"].emplace({
-			{ "comp_id", id },
+		db["chem_comp_bond"].emplace({ { "comp_id", id },
 			{ "atom_id_1", atom_id_1 },
 			{ "atom_id_2", atom_id_2 },
 			{ "value_order", value_order },
 			{ "pdbx_aromatic_flag", aromatic },
 			// TODO: fetch stereo_config info from chem_comp_chir
-			{ "pdbx_ordinal", ord++ }
-		});
+			{ "pdbx_ordinal", ord++ } });
 	}
 
 	db.emplace_back(rdb["pdbx_chem_comp_descriptor"]);
 
 	std::string formula;
-	for (bool first = true; const auto &[symbol, count]: formula_data)
+	for (bool first = true; const auto &[symbol, count] : formula_data)
 	{
 		if (std::exchange(first, false))
 			formula += ' ';
@@ -594,15 +582,13 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 	else
 		type = "NON-POLYMER";
 
-	db["chem_comp"].emplace({
-		{ "id", id },
+	db["chem_comp"].emplace({ { "id", id },
 		{ "name", name },
 		{ "type", type },
 		{ "formula", formula },
 		{ "pdbx_formal_charge", formal_charge },
 		{ "formula_weight", formula_weight },
-		{ "three_letter_code", three_letter_code }
-	});
+		{ "three_letter_code", three_letter_code } });
 
 	std::shared_lock lock(mMutex);
 
@@ -615,11 +601,16 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 
 std::unique_ptr<compound_factory> compound_factory::s_instance;
 thread_local std::unique_ptr<compound_factory> compound_factory::tl_instance;
-bool compound_factory::s_use_thread_local_instance;
+compound_factory_options compound_factory::s_options;
 
 void compound_factory::init(bool useThreadLocalInstanceOnly)
 {
-	s_use_thread_local_instance = useThreadLocalInstanceOnly;
+	init({ .use_thread_local_instance_only = useThreadLocalInstanceOnly });
+}
+
+void compound_factory::init(compound_factory_options options)
+{
+	s_options = options;
 }
 
 compound_factory::compound_factory()
@@ -638,7 +629,7 @@ compound_factory::~compound_factory()
 
 compound_factory &compound_factory::instance()
 {
-	if (s_use_thread_local_instance)
+	if (s_options.use_thread_local_instance_only)
 	{
 		if (not tl_instance)
 			tl_instance.reset(new compound_factory());
@@ -654,7 +645,7 @@ compound_factory &compound_factory::instance()
 
 void compound_factory::clear()
 {
-	if (s_use_thread_local_instance)
+	if (s_options.use_thread_local_instance_only)
 		tl_instance.reset(nullptr);
 	else
 		s_instance.reset();
@@ -732,7 +723,7 @@ bool compound_factory::is_peptide(std::string_view res_name) const
 	bool result = is_std_peptide(res_name);
 	if (not result and m_impl)
 	{
-		auto compound = const_cast<compound_factory&>(*this).create(res_name);
+		auto compound = const_cast<compound_factory &>(*this).create(res_name);
 		result = compound != nullptr and compound->is_peptide();
 	}
 	return result;
@@ -744,7 +735,7 @@ bool compound_factory::is_base(std::string_view res_name) const
 	bool result = is_std_base(res_name);
 	if (not result and m_impl)
 	{
-		auto compound = const_cast<compound_factory&>(*this).create(res_name);
+		auto compound = const_cast<compound_factory &>(*this).create(res_name);
 		result = compound != nullptr and compound->is_base();
 	}
 	return result;

src/condition.cpp

@@ -26,6 +26,7 @@
 
 #include "cif++/category.hpp"
 #include "cif++/condition.hpp"
+#include "cif++/validate.hpp"
 
 namespace cif
 {

src/datablock.cpp

@@ -25,6 +25,7 @@
  */
 
 #include "cif++/datablock.hpp"
+#include "cif++/validate.hpp"
 
 namespace cif
 {
@@ -41,30 +42,7 @@ datablock::datablock(const datablock &db)
 void datablock::load_dictionary()
 {
 	if (auto *audit_conform = get("audit_conform"); audit_conform and not audit_conform->empty())
-	{
-		std::string name = audit_conform->front().get<std::string>("dict_name");
-
-		if (name == "mmcif_pdbx_v50")
-			name = "mmcif_pdbx.dic"; // we had a bug here in libcifpp...
-
-		if (not name.empty())
-		{
-			try
-			{
-				load_dictionary(name);
-			}
-			catch (const std::exception &ex)
-			{
-				if (VERBOSE)
-					std::cerr << "Failed to load dictionary " << std::quoted(name) << ": " << ex.what() << '\n';
-			}
-		}
-	}
-}
-
-void datablock::load_dictionary(std::string_view name)
-{
-	set_validator(&validator_factory::instance()[name]);
+		set_validator(&validator_factory::instance().get(*audit_conform));
 }
 
 void datablock::set_validator(const validator *v)
@@ -100,40 +78,6 @@ bool datablock::is_valid() const
 	return result;
 }
 
-bool datablock::is_valid()
-{
-	if (m_validator == nullptr)
-		throw std::runtime_error("Validator not specified for datablock data_" + name());
-
-	bool result = true;
-	for (auto &cat : *this)
-		result = cat.is_valid() and result;
-	
-	// Add or remove the audit_conform block here.
-	if (result)
-	{
-		// If the dictionary declares an audit_conform category, put it in,
-		// but only if it does not exist already!
-
-		if (m_validator->get_validator_for_category("audit_conform") != nullptr)
-		{
-			auto &audit_conform = operator[]("audit_conform");
-
-			audit_conform.clear();
-			audit_conform.emplace({
-				// clang-format off
-				{ "dict_name", m_validator->name() },
-				{ "dict_version", m_validator->version() }
-				// clang-format on
-			});
-		}
-	}
-	else
-		erase(std::find_if(begin(), end(), [](category &cat) { return cat.name() == "audit_conform"; }), end());
-
-	return result;
-}
-
 bool datablock::validate_links() const
 {
 	bool result = true;
@@ -147,6 +91,45 @@ bool datablock::validate_links() const
 	return result;
 }
 
+bool datablock::strip()
+{
+	bool result = true;
+
+	// remove all categories that have no validator
+	erase(std::remove_if(begin(), end(), [](category &c) {
+		bool result = false;
+		if (c.get_cat_validator() == nullptr)
+		{
+			if (cif::VERBOSE > 0)
+				std::clog << "Dropping category " << c.name() << '\n';
+			result = true;
+		}
+		return result;
+	}), end());
+
+	// then strip the remaining categories
+	for (auto &cat : *this)
+		cat.strip();
+
+	// Add or remove the audit_conform block here.
+	if (is_valid())
+	{
+		// If the dictionary declares an audit_conform category, put it in,
+		// but only if it does not exist already!
+
+		if (auto audit_conform = get("audit_conform");
+			audit_conform != nullptr and m_validator->get_validator_for_category("audit_conform") != nullptr)
+		{
+			audit_conform->clear();
+			m_validator->fill_audit_conform(*audit_conform);
+		}
+	}
+	else
+		result = false;
+
+	return result;
+}
+
 // --------------------------------------------------------------------
 
 category &datablock::operator[](std::string_view name)
@@ -203,7 +186,7 @@ std::tuple<datablock::iterator, bool> datablock::emplace(std::string_view name)
 
 	if (is_new)
 	{
-		i = insert(end(), {name});
+		i = insert(end(), { name });
 		i->set_validator(m_validator, *this);
 	}
 
@@ -472,4 +455,4 @@ bool datablock::operator==(const datablock &rhs) const
 	return true;
 }
 
-} // namespace cif
+} // namespace cif

src/dictionary_parser.cpp

@@ -96,14 +96,10 @@ class dictionary_parser : public parser
 			link_items();
 
 		// store meta information
-		datablock::iterator info;
-		bool is_new;
-		std::tie(info, is_new) = m_datablock->emplace("dictionary");
-		if (not is_new and not info->empty())
+		if (auto dictionary = m_datablock->get("dictionary"); dictionary != nullptr and not dictionary->empty())
 		{
-			auto r = info->front();
-			m_validator.set_name(r["title"].as<std::string>());
-			m_validator.set_version(r["version"].as<std::string>());
+			const auto &[name, version] = dictionary->front().get<std::string,std::optional<std::string>>("title", "version");
+			m_validator.append_audit_conform(name, version);
 		}
 
 		m_datablock = savedDatablock;
@@ -252,7 +248,7 @@ class dictionary_parser : public parser
 
 				auto vi = find(ivs.begin(), ivs.end(), item_validator{ item_name });
 				if (vi == ivs.end())
-					ivs.push_back(item_validator{ item_name, iequals(mandatory, "yes"), tv, ess, defaultValue, nullptr, std::move(aliases) });
+					ivs.push_back(item_validator{ item_name, iequals(mandatory, "yes"), tv, ess, defaultValue, cat_name, std::move(aliases) });
 				else
 				{
 					// need to update the itemValidator?
@@ -350,7 +346,7 @@ class dictionary_parser : public parser
 			if (piv == nullptr)
 				error("in pdbx_item_linked_group_list, item '" + parent + "' is not specified");
 
-			key_type key{ piv->m_category->m_name, civ->m_category->m_name, link_group_id };
+			key_type key{ piv->m_category, civ->m_category, link_group_id };
 			if (not linkIndex.count(key))
 			{
 				linkIndex[key] = linkKeys.size();
@@ -378,7 +374,7 @@ class dictionary_parser : public parser
 				if (piv == nullptr)
 					error("in pdbx_item_linked_group_list, item '" + parent + "' is not specified");
 
-				key_type key{ piv->m_category->m_name, civ->m_category->m_name, 0 };
+				key_type key{ piv->m_category, civ->m_category, 0 };
 				if (not linkIndex.count(key))
 				{
 					linkIndex[key] = linkKeys.size();
@@ -477,18 +473,7 @@ class dictionary_parser : public parser
 
 // --------------------------------------------------------------------
 
-validator parse_dictionary(std::string_view name, std::istream &is)
-{
-	validator result(name);
-
-	file f;
-	dictionary_parser p(result, is, f);
-	p.load_dictionary();
-
-	return result;
-}
-
-void extend_dictionary(validator &v, std::istream &is)
+void parse_dictionary(validator &v, std::istream &is)
 {
 	file f;
 	dictionary_parser p(v, is, f);

src/file.cpp

@@ -65,42 +65,6 @@ bool file::validate_links() const
 	return result;
 }
 
-// void file::load_dictionary()
-// {
-// 	if (not empty())
-// 	{
-// 		auto *audit_conform = front().get("audit_conform");
-// 		if (audit_conform and not audit_conform->empty())
-// 		{
-// 			std::string name = audit_conform->front().get<std::string>("dict_name");
-
-// 			if (name == "mmcif_pdbx_v50")
-// 				name = "mmcif_pdbx.dic"; // we had a bug here in libcifpp...
-
-// 			if (not name.empty())
-// 			{
-// 				try
-// 				{
-// 					load_dictionary(name);
-// 				}
-// 				catch (const std::exception &ex)
-// 				{
-// 					if (VERBOSE)
-// 						std::cerr << "Failed to load dictionary " << std::quoted(name) << ": " << ex.what() << '\n';
-// 				}
-// 			}
-// 		}
-// 	}
-
-// 	// if (not m_validator)
-// 	// 	load_dictionary("mmcif_pdbx.dic");	// TODO: maybe incorrect? Perhaps improve?
-// }
-
-// void file::load_dictionary(std::string_view name)
-// {
-// 	set_validator(&validator_factory::instance()[name]);
-// }
-
 bool file::contains(std::string_view name) const
 {
 	return std::find_if(begin(), end(), [name](const datablock &db)
@@ -166,16 +130,6 @@ void file::load(const std::filesystem::path &p)
 	}
 }
 
-void file::load(const std::filesystem::path &p, std::string_view dict)
-{
-	load(p, validator_factory::instance().operator[](dict));
-}
-
-void file::load(std::istream &is, std::string_view dict)
-{
-	load(is, validator_factory::instance().operator[](dict));
-}
-
 void file::load(const std::filesystem::path &p, const validator &v)
 {
 	gzio::ifstream in(p);
@@ -221,4 +175,4 @@ void file::save(std::ostream &os) const
 		db.write(os);
 }
 
-} // namespace cif
+} // namespace cif

src/model.cpp

@@ -102,18 +102,18 @@ void atom::atom_impl::set_property(const std::string_view name, const std::strin
 	r.assign(name, value, true, true);
 }
 
-// int atom::atom_impl::compare(const atom_impl &b) const
-// {
-// 	int d = m_asym_id.compare(b.m_asym_id);
-// 	if (d == 0)
-// 		d = m_seq_id - b.m_seq_id;
-// 	if (d == 0)
-// 		d = m_auth_seq_id.compare(b.m_auth_seq_id);
-// 	if (d == 0)
-// 		d = mAtom_id.compare(b.mAtom_id);
+int atom::atom_impl::compare(const atom_impl &b) const
+{
+	int d = get_property("label_asym_id").compare(b.get_property("label_asym_id"));
+	if (d == 0)
+		d = get_property_int("label_seq_id") - b.get_property_int("label_seq_id");
+	if (d == 0)
+		d = get_property_int("auth_seq_id") - b.get_property_int("auth_seq_id");
+	if (d == 0)
+		d = get_property("label_atom_id").compare(b.get_property("label_atom_id"));
 
-// 	return d;
-// }
+	return d;
+}
 
 // bool atom::atom_impl::getAnisoU(float anisou[6]) const
 // {
@@ -149,145 +149,6 @@ int atom::atom_impl::get_charge() const
 	return formalCharge.value_or(0);
 }
 
-// const Compound *atom::atom_impl::compound() const
-// {
-// 	if (mCompound == nullptr)
-// 	{
-// 		std::string compID = get_property("label_comp_id");
-
-// 		mCompound = compound_factory::instance().create(compID);
-// 	}
-
-// 	return mCompound;
-// }
-
-// const std::string atom::atom_impl::get_property(const std::string_view name) const
-// {
-// 	for (auto &&[item_name, ref] : mCachedRefs)
-// 	{
-// 		if (item_name == name)
-// 			return ref.as<std::string>();
-// 	}
-
-// 	mCachedRefs.emplace_back(name, const_cast<Row &>(mRow)[name]);
-// 	return std::get<1>(mCachedRefs.back()).as<std::string>();
-// }
-
-// void atom::atom_impl::set_property(const std::string_view name, const std::string &value)
-// {
-// 	for (auto &&[item_name, ref] : mCachedRefs)
-// 	{
-// 		if (item_name != name)
-// 			continue;
-
-// 		ref = value;
-// 		return;
-// 	}
-
-// 	mCachedRefs.emplace_back(name, mRow[name]);
-// 	std::get<1>(mCachedRefs.back()) = value;
-// }
-
-// const Row atom::getRowAniso() const
-// {
-// 	auto &db = m_impl->m_db;
-// 	auto cat = db.get("atom_site_anisotrop");
-// 	if (not cat)
-// 		return {};
-// 	else
-// 		return cat->find1(key("id") == m_impl->m_id);
-// }
-
-// float atom::uIso() const
-// {
-// 	float result;
-
-// 	if (not get_property<std::string>("U_iso_or_equiv").empty())
-// 		result = get_property<float>("U_iso_or_equiv");
-// 	else if (not get_property<std::string>("B_iso_or_equiv").empty())
-// 		result = get_property<float>("B_iso_or_equiv") / static_cast<float>(8 * kPI * kPI);
-// 	else
-// 		throw std::runtime_error("Missing B_iso or U_iso");
-
-// 	return result;
-// }
-
-// const Compound &atom::compound() const
-// {
-// 	auto result = impl().compound();
-
-// 	if (result == nullptr)
-// 	{
-// 		if (VERBOSE > 0)
-// 			std::cerr << "Compound not found: '" << get_property<std::string>("label_comp_id") << '\'' << '\n';
-
-// 		throw std::runtime_error("no compound");
-// 	}
-
-// 	return *result;
-// }
-
-// std::string atom::labelEntityID() const
-// {
-// 	return get_property<std::string>("label_entity_id");
-// }
-
-// std::string atom::authAtom_id() const
-// {
-// 	return get_property<std::string>("auth_atom_id");
-// }
-
-// std::string atom::authCompID() const
-// {
-// 	return get_property<std::string>("auth_comp_id");
-// }
-
-// std::string atom::get_auth_asym_id() const
-// {
-// 	return get_property<std::string>("auth_asym_id");
-// }
-
-// std::string atom::get_pdb_ins_code() const
-// {
-// 	return get_property<std::string>("pdbx_PDB_ins_code");
-// }
-
-// std::string atom::pdbxAuthAltID() const
-// {
-// 	return get_property<std::string>("pdbx_auth_alt_id");
-// }
-
-// void atom::translate(point t)
-// {
-// 	auto loc = location();
-// 	loc += t;
-// 	location(loc);
-// }
-
-// void atom::rotate(quaternion q)
-// {
-// 	auto loc = location();
-// 	loc.rotate(q);
-// 	location(loc);
-// }
-
-// void atom::translate_and_rotate(point t, quaternion q)
-// {
-// 	auto loc = location();
-// 	loc += t;
-// 	loc.rotate(q);
-// 	location(loc);
-// }
-
-// void atom::translate_rotate_and_translate(point t1, quaternion q, point t2)
-// {
-// 	auto loc = location();
-// 	loc += t1;
-// 	loc.rotate(q);
-// 	loc += t2;
-// 	location(loc);
-// }
-
 std::ostream &operator<<(std::ostream &os, const atom &atom)
 {
 	if (atom.is_water())
@@ -319,8 +180,8 @@ residue::residue(structure &structure, const std::vector<atom> &atoms)
 	m_compound_id = a.get_label_comp_id();
 	m_asym_id = a.get_label_asym_id();
 	m_seq_id = a.get_label_seq_id();
-	m_auth_asym_id = a.get_auth_asym_id();
-	m_auth_seq_id = a.get_auth_seq_id();
+	m_pdb_strand_id = a.get_auth_asym_id();
+	m_pdb_seq_num = a.get_auth_seq_id();
 	m_pdb_ins_code = a.get_pdb_ins_code();
 
 	for (auto atom : atoms)
@@ -371,11 +232,12 @@ atom residue::create_new_atom(atom_type inType, const std::string &inAtomID, poi
 		{ "label_alt_id", "." },
 		{ "label_comp_id", m_compound_id },
 		{ "label_seq_id", m_seq_id },
-		{ "auth_asym_id", m_auth_asym_id },
+		{ "auth_asym_id", m_pdb_strand_id },
 		{ "auth_atom_id", inAtomID },
 		{ "auth_comp_id", m_compound_id },
-		{ "auth_seq_id", m_auth_seq_id },
+		{ "auth_seq_id", m_pdb_seq_num },
 		{ "occupancy", 1.0f, 2 },
+		{ "B_iso_or_equiv", 20.0f },
 		{ "pdbx_PDB_model_num", m_structure->get_model_nr() },
 	});
 
@@ -450,6 +312,28 @@ atom residue::get_atom_by_atom_id(const std::string &atom_id) const
 	return result;
 }
 
+atom residue::get_atom_by_atom_id(const std::string &atomID, const std::string &altID) const
+{
+	if (altID.empty())
+		return get_atom_by_atom_id(atomID);
+
+	atom result;
+
+	for (auto &a : m_atoms)
+	{
+		if (auto a_alt_id = a.get_label_alt_id(); a.get_label_atom_id() == atomID and (a_alt_id.empty() or a_alt_id == altID))
+		{
+			result = a;
+			break;
+		}
+	}
+
+	if (not result and VERBOSE > 1)
+		std::cerr << "atom with atom_id " << atomID << " and alt_id " << altID << " not found in residue " << m_asym_id << ':' << m_seq_id << '\n';
+
+	return result;
+}
+
 // residue is a single entity if the atoms for the asym with m_asym_id is equal
 // to the number of atoms in this residue...  hope this is correct....
 bool residue::is_entity() const
@@ -518,8 +402,8 @@ std::ostream &operator<<(std::ostream &os, const residue &res)
 {
 	os << res.get_compound_id() << ' ' << res.get_asym_id() << ':' << res.get_seq_id();
 
-	if (res.get_auth_asym_id() != res.get_asym_id() or res.get_auth_seq_id() != std::to_string(res.get_seq_id()))
-		os << " [" << res.get_auth_asym_id() << ':' << res.get_auth_seq_id() << ']';
+	if (res.get_pdb_strand_id() != res.get_asym_id() or res.get_pdb_seq_num() != std::to_string(res.get_seq_id()))
+		os << " [" << res.get_pdb_strand_id() << ':' << res.get_pdb_seq_num() << ']';
 
 	return os;
 }
@@ -528,7 +412,7 @@ std::ostream &operator<<(std::ostream &os, const residue &res)
 // monomer
 
 monomer::monomer(const polymer &polymer, std::size_t index, int seqID, const std::string &authSeqID, const std::string &pdbInsCode, const std::string &compoundID)
-	: residue(*polymer.get_structure(), compoundID, polymer.get_asym_id(), seqID, polymer.get_auth_asym_id(), authSeqID, pdbInsCode)
+	: residue(*polymer.get_structure(), compoundID, polymer.get_asym_id(), seqID, polymer.get_pdb_strand_id(), authSeqID, pdbInsCode)
 	, m_polymer(&polymer)
 	, m_index(index)
 {
@@ -562,6 +446,16 @@ bool monomer::is_last_in_chain() const
 	return m_index + 1 == m_polymer->size();
 }
 
+const monomer &monomer::prev() const
+{
+	return m_polymer->at(m_index - 1);
+}
+
+const monomer &monomer::next() const
+{
+	return m_polymer->at(m_index + 1);
+}
+
 bool monomer::has_alpha() const
 {
 	return m_index >= 1 and m_index + 2 < m_polymer->size();
@@ -578,7 +472,7 @@ float monomer::phi() const
 
 	if (m_index > 0)
 	{
-		auto &prev = m_polymer->operator[](m_index - 1);
+		auto &prev = m_polymer->at(m_index - 1);
 		if (prev.m_seq_id + 1 == m_seq_id)
 		{
 			auto a1 = prev.C();
@@ -600,7 +494,7 @@ float monomer::psi() const
 
 	if (m_index + 1 < m_polymer->size())
 	{
-		auto &next = m_polymer->operator[](m_index + 1);
+		auto &next = m_polymer->at(m_index + 1);
 		if (m_seq_id + 1 == next.m_seq_id)
 		{
 			auto a1 = N();
@@ -624,9 +518,9 @@ float monomer::alpha() const
 	{
 		if (m_index >= 1 and m_index + 2 < m_polymer->size())
 		{
-			auto &prev = m_polymer->operator[](m_index - 1);
-			auto &next = m_polymer->operator[](m_index + 1);
-			auto &nextNext = m_polymer->operator[](m_index + 2);
+			auto &prev = m_polymer->at(m_index - 1);
+			auto &next = m_polymer->at(m_index + 1);
+			auto &nextNext = m_polymer->at(m_index + 2);
 
 			result = static_cast<float>(dihedral_angle(prev.CAlpha().get_location(), CAlpha().get_location(), next.CAlpha().get_location(), nextNext.CAlpha().get_location()));
 		}
@@ -648,8 +542,8 @@ float monomer::kappa() const
 	{
 		if (m_index >= 2 and m_index + 2 < m_polymer->size())
 		{
-			auto &prevPrev = m_polymer->operator[](m_index - 2);
-			auto &nextNext = m_polymer->operator[](m_index + 2);
+			auto &prevPrev = m_polymer->at(m_index - 2);
+			auto &nextNext = m_polymer->at(m_index + 2);
 
 			if (prevPrev.m_seq_id + 4 == nextNext.m_seq_id)
 			{
@@ -677,7 +571,7 @@ float monomer::tco() const
 	{
 		if (m_index > 0)
 		{
-			auto &prev = m_polymer->operator[](m_index - 1);
+			auto &prev = m_polymer->at(m_index - 1);
 			if (prev.m_seq_id + 1 == m_seq_id)
 				result = static_cast<float>(cosinus_angle(C().get_location(), O().get_location(), prev.C().get_location(), prev.O().get_location()));
 		}
@@ -699,7 +593,7 @@ float monomer::omega() const
 	try
 	{
 		if (not is_last_in_chain())
-			result = omega(*this, m_polymer->operator[](m_index + 1));
+			result = omega(*this, m_polymer->at(m_index + 1));
 	}
 	catch (const std::exception &ex)
 	{
@@ -795,7 +689,7 @@ bool monomer::is_cis() const
 
 	if (m_index + 1 < m_polymer->size())
 	{
-		auto &next = m_polymer->operator[](m_index + 1);
+		auto &next = m_polymer->at(m_index + 1);
 
 		result = monomer::is_cis(*this, next);
 	}
@@ -937,7 +831,7 @@ polymer::polymer(structure &s, const std::string &entityID, const std::string &a
 	: m_structure(const_cast<structure *>(&s))
 	, m_entity_id(entityID)
 	, m_asym_id(asym_id)
-	, m_auth_asym_id(auth_asym_id)
+	, m_pdb_strand_id(auth_asym_id)
 {
 	using namespace cif::literals;
 
@@ -949,12 +843,8 @@ polymer::polymer(structure &s, const std::string &entityID, const std::string &a
 	for (auto r : poly_seq_scheme.find("asym_id"_key == asym_id))
 	{
 		int seqID;
-		std::optional<int> pdbSeqNum;
-		std::string compoundID, authSeqID, pdbInsCode;
-		cif::tie(seqID, authSeqID, compoundID, pdbInsCode, pdbSeqNum) = r.get("seq_id", "auth_seq_num", "mon_id", "pdb_ins_code", "pdb_seq_num");
-
-		if (authSeqID.empty() and pdbSeqNum.has_value())
-			authSeqID = std::to_string(*pdbSeqNum);
+		std::string compoundID, pdbSeqNum, pdbInsCode;
+		cif::tie(seqID, pdbSeqNum, compoundID, pdbInsCode) = r.get("seq_id", "pdb_seq_num", "mon_id", "pdb_ins_code");
 
 		std::size_t index = size();
 
@@ -962,11 +852,11 @@ polymer::polymer(structure &s, const std::string &entityID, const std::string &a
 		if (not ix.count(seqID))
 		{
 			ix[seqID] = index;
-			emplace_back(*this, index, seqID, authSeqID, pdbInsCode, compoundID);
+			emplace_back(*this, index, seqID, pdbSeqNum, pdbInsCode, compoundID);
 		}
 		else if (VERBOSE > 0)
 		{
-			monomer m{ *this, index, seqID, authSeqID, pdbInsCode, compoundID };
+			monomer m{ *this, index, seqID, pdbSeqNum, pdbInsCode, compoundID };
 			std::cerr << "Dropping alternate residue " << m << '\n';
 		}
 	}
@@ -1106,7 +996,7 @@ cif::mm::atom sugar::add_atom(row_initializer atom_info)
 	atom_info.set_value({ "label_alt_id", "." });
 	atom_info.set_value({ "auth_asym_id", m_branch->get_asym_id() });
 	atom_info.set_value({ "auth_comp_id", m_compound_id });
-	atom_info.set_value({ "auth_seq_id", m_auth_seq_id });
+	atom_info.set_value({ "auth_seq_id", m_pdb_seq_num });
 	atom_info.set_value({ "occupancy", 1.0, 2 });
 	atom_info.set_value({ "B_iso_or_equiv", 30.0, 2 });
 	atom_info.set_value({ "pdbx_PDB_model_num", 1 });
@@ -1222,12 +1112,12 @@ sugar &branch::construct_sugar(const std::string &compound_id)
 		{ "mon_id", result.get_compound_id() },
 
 		{ "pdb_asym_id", result.get_asym_id() },
-		{ "pdb_seq_num", result.num() },
+		{ "pdb_seq_num", result.get_pdb_seq_num() },
 		{ "pdb_mon_id", result.get_compound_id() },
 
-		{ "auth_asym_id", result.get_auth_asym_id() },
+		{ "auth_asym_id", result.get_pdb_strand_id() },
 		{ "auth_mon_id", result.get_compound_id() },
-		{ "auth_seq_num", result.get_auth_seq_id() },
+		{ "auth_seq_num", result.get_pdb_seq_num() },
 
 		{ "hetero", "n" } });
 
@@ -1270,7 +1160,7 @@ std::string branch::name(const sugar &s) const
 
 	for (auto &sn : *this)
 	{
-		if (not sn.get_link() or sn.get_link().get_auth_seq_id() != s.get_auth_seq_id())
+		if (not sn.get_link() or sn.get_link().get_auth_seq_id() != s.get_pdb_seq_num())
 			continue;
 
 		auto n = name(sn) + "-(1-" + sn.get_link().get_label_atom_id().substr(1) + ')';
@@ -1297,16 +1187,19 @@ float branch::weight() const
 // --------------------------------------------------------------------
 //	structure
 
-structure::structure(file &p, std::size_t modelNr, StructureOpenOptions options)
+structure::structure(file &p, std::size_t modelNr, structure_open_options options)
 	: structure(p.front(), modelNr, options)
 {
 }
 
-structure::structure(datablock &db, std::size_t modelNr, StructureOpenOptions options)
+structure::structure(datablock &db, std::size_t modelNr, structure_open_options options)
 	: m_db(db)
 	, m_model_nr(modelNr)
 {
-	auto &atomCat = db["atom_site"];
+	if (db.get_validator() == nullptr)
+		db.load_dictionary();
+
+	auto &atom_site = db["atom_site"];
 
 	load_atoms_for_model(options);
 
@@ -1314,7 +1207,7 @@ structure::structure(datablock &db, std::size_t modelNr, StructureOpenOptions op
 	if (m_atoms.empty() and m_model_nr == 1)
 	{
 		std::optional<std::size_t> model_nr;
-		cif::tie(model_nr) = atomCat.front().get("pdbx_PDB_model_num");
+		cif::tie(model_nr) = atom_site.front().get("pdbx_PDB_model_num");
 		if (model_nr and *model_nr != m_model_nr)
 		{
 			if (VERBOSE > 0)
@@ -1333,42 +1226,133 @@ structure::structure(datablock &db, std::size_t modelNr, StructureOpenOptions op
 		load_data();
 }
 
-void structure::load_atoms_for_model(StructureOpenOptions options)
+void structure::load_atoms_for_model(structure_open_options options)
 {
 	using namespace literals;
 
-	auto &atomCat = m_db["atom_site"];
+	auto &atom_site = m_db["atom_site"];
 
 	condition c = "pdbx_PDB_model_num"_key == null or "pdbx_PDB_model_num"_key == m_model_nr;
-	if (options bitand StructureOpenOptions::SkipHydrogen)
-		c = std::move(c) and ("type_symbol"_key != "H" and "type_symbol"_key != "D");
 
-	for (auto id : atomCat.find<std::string>(std::move(c), "id"))
-		emplace_atom(std::make_shared<atom::atom_impl>(m_db, id));
+	if (options.skip_hydrogen)
+		c = std::move(c) and (cif::key("type_symbol") != "H" and cif::key("type_symbol") != "D");
+
+	if (options.skip_water)
+		c = std::move(c) and (cif::key("auth_comp_id") != "HOH" and cif::key("auth_comp_id") != "H20" and cif::key("auth_comp_id") != "WAT");
+
+	if (options.skip_hetatom)
+	{
+		if (options.skip_water)
+			c = std::move(c) and cif::key("group_PDB") != "HETATM";
+		else
+			c = std::move(c) and (cif::key("group_PDB") != "HETATM" or (cif::key("auth_comp_id") == "HOH" or cif::key("auth_comp_id") == "H20" or cif::key("auth_comp_id") == "WAT"));
+	}
+
+	if (options.min_b_factor.has_value())
+		c = std::move(c) and cif::key("B_iso_or_equiv") >= *options.min_b_factor;
+
+	if (options.max_b_factor.has_value())
+		c = std::move(c) and cif::key("B_iso_or_equiv") <= *options.max_b_factor;
+
+	if (not options.asyms.empty())
+	{
+		condition tmp_c;
+		for (auto asym_id : options.asyms)
+			tmp_c = std::move(tmp_c) or cif::key("label_asym_id") == asym_id;
+		c = std::move(c) and std::move(tmp_c);
+	}
+
+	if (options.occupancy_mode == occupancy_policy::ALL)
+	{
+		for (auto id : atom_site.find<std::string>(std::move(c), "id"))
+			emplace_atom(std::make_shared<atom::atom_impl>(m_db, id));
+	}
+	else if (options.occupancy_mode == occupancy_policy::UNOCCUPIED)
+	{
+		for (auto id : atom_site.find<std::string>(std::move(c), "id"))
+		{
+			auto a = std::make_shared<atom::atom_impl>(m_db, id);
+			if (a->get_property_float("occupancy") > 0)
+				continue;
+			emplace_atom(a);
+		}
+	}
+	else
+	{
+		std::vector<cif::mm::atom> atoms;
+		std::map<std::tuple<std::string,int>, std::map<std::string, float>> alts;
+	
+		for (auto id : atom_site.find<std::string>(std::move(c), "id"))
+		{
+			auto a = atoms.emplace_back(std::make_shared<atom::atom_impl>(m_db, id));
+	
+			if (a.is_alternate())
+			{
+				auto key = std::make_tuple(a.get_label_asym_id(), a.get_label_seq_id());
+				auto alt_id = a.get_label_alt_id();
+	
+				if (auto i = alts.find(key); i != alts.end())
+					i->second[alt_id] += a.get_occupancy();
+				else
+					alts[key][alt_id] = a.get_occupancy();
+			}
+		}
+	
+		for (auto &&[key, value] : alts)
+		{
+			const auto &[asym_id, seq_id] = key;
+	
+			// select highest occupancy for this residue's alternates
+			std::string alt_id;
+			float occupancy = options.occupancy_mode == occupancy_policy::MAX ? 0.f : std::numeric_limits<float>::max();
+			for (const auto &[alt_key, alt_value] : value)
+			{
+				if (options.occupancy_mode == occupancy_policy::MAX)
+				{
+					if (occupancy < alt_value)
+					{
+						alt_id = alt_key;
+						occupancy = alt_value;
+					}
+				}
+				else
+				{
+					if (occupancy > alt_value)
+					{
+						alt_id = alt_key;
+						occupancy = alt_value;
+					}
+				}
+			}
+	
+			value.clear();
+			value.emplace(alt_id, occupancy);
+		}
+	
+		for (auto a : atoms)
+		{
+			if (a.is_alternate())
+			{
+				auto key = std::make_tuple(a.get_label_asym_id(), a.get_label_seq_id());
+
+				if (alts[key].contains(a.get_label_alt_id()))
+					emplace_atom(a);
+			}
+			else
+				emplace_atom(a);
+	
+		}
+	}
+
 }
 
-// structure::structure(const structure &s)
-// 	: m_db(s.m_db)
-// 	, m_model_nr(s.m_model_nr)
-// {
-// 	m_atoms.reserve(s.m_atoms.size());
-// 	for (auto &atom : s.m_atoms)
-// 		emplace_atom(atom.clone());
-
-// 	load_data();
-// }
-
-// structure::~structure()
-// {
-// }
-
 void structure::load_data()
 {
 	auto &polySeqScheme = m_db["pdbx_poly_seq_scheme"];
 
 	for (const auto &[asym_id, auth_asym_id, entityID] : polySeqScheme.rows<std::string, std::string, std::string>("asym_id", "pdb_strand_id", "entity_id"))
 	{
-		if (m_polymers.empty() or m_polymers.back().get_asym_id() != asym_id or m_polymers.back().get_entity_id() != entityID)
+		if (m_polymers.empty() or m_polymers.back().get_asym_id() != asym_id)
 			m_polymers.emplace_back(*this, entityID, asym_id, auth_asym_id);
 	}
 
@@ -1394,18 +1378,18 @@ void structure::load_data()
 	for (auto &poly : m_polymers)
 	{
 		for (auto &res : poly)
-			resMap[{ res.get_asym_id(), res.get_seq_id(), res.get_auth_seq_id() }] = &res;
+			resMap[{ res.get_asym_id(), res.get_seq_id(), res.get_pdb_seq_num() }] = &res;
 	}
 
 	for (auto &res : m_non_polymers)
-		resMap[{ res.get_asym_id(), res.get_seq_id(), res.get_auth_seq_id() }] = &res;
+		resMap[{ res.get_asym_id(), res.get_seq_id(), res.get_pdb_seq_num() }] = &res;
 
 	std::set<std::string> sugars;
 	for (auto &branch : m_branches)
 	{
 		for (auto &sugar : branch)
 		{
-			resMap[{ sugar.get_asym_id(), sugar.get_seq_id(), sugar.get_auth_seq_id() }] = &sugar;
+			resMap[{ sugar.get_asym_id(), sugar.get_seq_id(), sugar.get_pdb_seq_num() }] = &sugar;
 			sugars.insert(sugar.get_compound_id());
 		}
 	}
@@ -1480,30 +1464,6 @@ EntityType structure::get_entity_type_for_asym_id(const std::string asym_id) con
 	return get_entity_type_for_entity_id(entityID);
 }
 
-// std::vector<atom> structure::waters() const
-// {
-// 	using namespace literals;
-
-// 	std::vector<atom> result;
-
-// 	auto &db = datablock();
-
-// 	// Get the entity id for water. Watch out, structure may not have water at all
-// 	auto &entityCat = db["entity"];
-// 	for (const auto &[waterEntityID] : entityCat.find<std::string>("type"_key == "water", "id"))
-// 	{
-// 		for (auto &a : m_atoms)
-// 		{
-// 			if (a.get_property("label_entity_id") == waterEntityID)
-// 				result.push_back(a);
-// 		}
-
-// 		break;
-// 	}
-
-// 	return result;
-// }
-
 bool structure::has_atom_id(const std::string &id) const
 {
 	assert(m_atoms.size() == m_atom_index.size());
@@ -1652,7 +1612,7 @@ residue &structure::get_residue(const std::string &asym_id, int seqID, const std
 	{
 		for (auto &res : m_non_polymers)
 		{
-			if (res.get_asym_id() == asym_id and (authSeqID.empty() or res.get_auth_seq_id() == authSeqID))
+			if (res.get_asym_id() == asym_id and (authSeqID.empty() or res.get_pdb_seq_num() == authSeqID))
 				return res;
 		}
 	}
@@ -1676,7 +1636,7 @@ residue &structure::get_residue(const std::string &asym_id, int seqID, const std
 
 		for (auto &sugar : branch)
 		{
-			if (sugar.get_asym_id() == asym_id and sugar.get_auth_seq_id() == authSeqID)
+			if (sugar.get_asym_id() == asym_id and sugar.get_pdb_seq_num() == authSeqID)
 				return sugar;
 		}
 	}
@@ -1698,7 +1658,7 @@ residue &structure::get_residue(const std::string &asym_id, const std::string &c
 	{
 		for (auto &res : m_non_polymers)
 		{
-			if (res.get_asym_id() == asym_id and res.get_auth_seq_id() == authSeqID and res.get_compound_id() == compID)
+			if (res.get_asym_id() == asym_id and res.get_pdb_seq_num() == authSeqID and res.get_compound_id() == compID)
 				return res;
 		}
 	}
@@ -1722,7 +1682,7 @@ residue &structure::get_residue(const std::string &asym_id, const std::string &c
 
 		for (auto &sugar : branch)
 		{
-			if (sugar.get_asym_id() == asym_id and sugar.get_auth_seq_id() == authSeqID and sugar.get_compound_id() == compID)
+			if (sugar.get_asym_id() == asym_id and sugar.get_pdb_seq_num() == authSeqID and sugar.get_compound_id() == compID)
 				return sugar;
 		}
 	}
@@ -2072,7 +2032,7 @@ void structure::remove_residue(const std::string &asym_id, int seq_id, const std
 	{
 		for (auto &res : m_non_polymers)
 		{
-			if (res.get_asym_id() == asym_id and (auth_seq_id.empty() or res.get_auth_seq_id() == auth_seq_id))
+			if (res.get_asym_id() == asym_id and (auth_seq_id.empty() or res.get_pdb_seq_num() == auth_seq_id))
 			{
 				remove_residue(res);
 				return;
@@ -2102,7 +2062,7 @@ void structure::remove_residue(const std::string &asym_id, int seq_id, const std
 
 		for (auto &sugar : branch)
 		{
-			if (sugar.get_asym_id() == asym_id and sugar.get_auth_seq_id() == auth_seq_id)
+			if (sugar.get_asym_id() == asym_id and sugar.get_pdb_seq_num() == auth_seq_id)
 			{
 				remove_residue(sugar);
 				return;
@@ -2235,7 +2195,7 @@ void structure::remove_sugar(sugar &s)
 				// TODO: need fix, collect from nag_atoms?
 				{ "auth_asym_id", asym_id },
 				{ "auth_mon_id", sugar.get_compound_id() },
-				{ "auth_seq_num", sugar.get_auth_seq_id() },
+				{ "auth_seq_num", sugar.get_pdb_seq_num() },
 
 				{ "hetero", "n" } });
 		}
@@ -2321,8 +2281,8 @@ std::string structure::create_non_poly(const std::string &entity_id, const std::
 		{ "entity_id", entity_id },
 		{ "mon_id", comp_id },
 		{ "ndb_seq_num", ndb_nr },
-		{ "pdb_seq_num", res.get_auth_seq_id() },
-		{ "auth_seq_num", res.get_auth_seq_id() },
+		{ "pdb_seq_num", res.get_pdb_seq_num() },
+		{ "auth_seq_num", res.get_pdb_seq_num() },
 		{ "pdb_mon_id", comp_id },
 		{ "auth_mon_id", comp_id },
 		{ "pdb_strand_id", asym_id },
@@ -2382,8 +2342,8 @@ std::string structure::create_non_poly(const std::string &entity_id, std::vector
 		{ "entity_id", entity_id },
 		{ "mon_id", comp_id },
 		{ "ndb_seq_num", ndb_nr },
-		{ "pdb_seq_num", res.get_auth_seq_id() },
-		{ "auth_seq_num", res.get_auth_seq_id() },
+		{ "pdb_seq_num", res.get_pdb_seq_num() },
+		{ "auth_seq_num", res.get_pdb_seq_num() },
 		{ "pdb_mon_id", comp_id },
 		{ "auth_mon_id", comp_id },
 		{ "pdb_strand_id", asym_id },
@@ -2697,11 +2657,11 @@ std::string structure::create_entity_for_branch(branch &branch)
 
 			pdbx_entity_branch_link.emplace({ { "link_id", pdbx_entity_branch_link.get_unique_id("") },
 				{ "entity_id", entityID },
-				{ "entity_branch_list_num_1", s1.get_auth_seq_id() },
+				{ "entity_branch_list_num_1", s1.get_pdb_seq_num() },
 				{ "comp_id_1", s1.get_compound_id() },
 				{ "atom_id_1", l1.get_label_atom_id() },
 				{ "leaving_atom_id_1", "O1" },
-				{ "entity_branch_list_num_2", s2.get_auth_seq_id() },
+				{ "entity_branch_list_num_2", s2.get_pdb_seq_num() },
 				{ "comp_id_2", s2.get_compound_id() },
 				{ "atom_id_2", l2.get_label_atom_id() },
 				{ "leaving_atom_id_2", "H" + l2.get_label_atom_id() },
@@ -2714,9 +2674,12 @@ std::string structure::create_entity_for_branch(branch &branch)
 
 void structure::cleanup_empty_categories()
 {
+
 	using namespace literals;
 
 	auto &atomSite = m_db["atom_site"];
+	auto &pdbxPolySeqScheme = m_db["pdbx_poly_seq_scheme"];
+	auto &entityPolySeq = m_db["entity_poly_seq"];
 
 	// Remove chem_comp's for which there are no atoms at all
 	auto &chem_comp = m_db["chem_comp"];
@@ -2725,8 +2688,12 @@ void structure::cleanup_empty_categories()
 	for (auto chemComp : chem_comp)
 	{
 		std::string compID = chemComp["id"].as<std::string>();
-		if (atomSite.contains("label_comp_id"_key == compID or "auth_comp_id"_key == compID))
+		if (atomSite.contains("label_comp_id"_key == compID or "auth_comp_id"_key == compID) or
+			pdbxPolySeqScheme.contains("mon_id"_key == compID or "auth_mon_id"_key == compID or "pdb_mon_id"_key == compID) or
+			entityPolySeq.contains("mon_id"_key == compID))
+		{
 			continue;
+		}
 
 		obsoleteChemComps.push_back(chemComp);
 	}
@@ -2905,6 +2872,14 @@ static int compare_numbers(std::string_view a, std::string_view b)
 	return result;
 }
 
+int compare_cif_id(const std::string &a, const std::string &b)
+{
+	int d = a.length() - b.length();
+	if (d == 0)
+		d = a.compare(b);
+	return d;
+}
+
 void structure::reorder_atoms()
 {
 	auto &atom_site = m_db["atom_site"];
@@ -2916,7 +2891,7 @@ void structure::reorder_atoms()
 			// First by model number
 			d = a.get<int>("pdbx_PDB_model_num") - b.get<int>("pdbx_PDB_model_num");
 			if (d == 0)
-				d = a.get<std::string>("label_asym_id").compare(b.get<std::string>("label_asym_id"));
+				d = compare_cif_id(a.get<std::string>("label_asym_id"), b.get<std::string>("label_asym_id"));
 			if (d == 0)
 			{
 				auto na = a.get<std::optional<int>>("label_seq_id");

src/pdb/pdb2cif.cpp

@@ -1962,7 +1962,7 @@ void PDBFileParser::ParseRemarks()
 			switch (remarkNr)
 			{
 				case 1:
-					while (mRec->is("REMARK   1"))
+					while (mRec->is("REMARK   1") or mRec->is("REMARK 001"))
 					{
 						if (mRec->mVlen > 15 and vS(12, 20) == "REFERENCE")
 						{
@@ -3334,15 +3334,9 @@ void PDBFileParser::ParseRemark350()
 
 						std::string type = mat == std::vector<double>{ 1, 0, 0, 0, 1, 0, 0, 0, 1 } and vec == std::vector<double>{ 0, 0, 0 } ? "identity operation" : "crystal symmetry operation";
 
-						// if (type == "identity operation")
-						// {
-
-						// }
-						// else
-						try
-						{
-							// clang-format off
-							getCategory("pdbx_struct_oper_list")->emplace({
+						auto pdbx_struct_oper_list = getCategory("pdbx_struct_oper_list");
+						if (not pdbx_struct_oper_list->contains(cif::key("id") == operID))
+							getCategory("pdbx_struct_oper_list")->emplace({ // clang-format off
 								{ "id", operID },
 								{ "type", type },
 								// { "name", "" },
@@ -3360,12 +3354,7 @@ void PDBFileParser::ParseRemark350()
 								{ "matrix[3][3]", cif::format("%12.10f", mat[8]).str() },
 								{ "vector[3]", cif::format("%12.10f", vec[2]).str() }
 							});
-							// clang-format on
-						}
-						catch (duplicate_key_error &ex)
-						{
-							// so what?
-						}
+																			// clang-format on
 
 						mat.clear();
 						vec.clear();
@@ -4300,6 +4289,8 @@ void PDBFileParser::ConstructEntities()
 			type = "polypeptide(L)";
 		else if (mightBeDNA and not mightBePolyPeptide)
 			type = "polyribonucleotide";
+		else
+			type = "other";
 
 		// clang-format off
 		getCategory("entity_poly")->emplace({
@@ -4505,7 +4496,7 @@ void PDBFileParser::ConstructEntities()
 	int modResID = 1;
 	std::set<std::string> modResSet;
 	for (auto rec = FindRecord("MODRES"); rec != nullptr and rec->is("MODRES");
-		 rec = rec->mNext)                     //	 1 -  6        Record name   "MODRES"
+		rec = rec->mNext)                      //	 1 -  6        Record name   "MODRES"
 	{                                          //	 8 - 11        IDcode        idCode      ID code of this datablock.
 		std::string resName = rec->vS(13, 15); //	13 - 15        Residue name  resName     Residue name used in this datablock.
 		char chainID = rec->vC(17);            //	17             Character     chainID     Chain identifier.
@@ -5627,7 +5618,7 @@ void PDBFileParser::ParseCoordinateTransformation()
 			igiven = vC(60) == '1';   //	60             Integer       iGiven        1 if coordinates for the  representations
 			                          //	                                           which  are approximately related by the
 			GetNextRecord();          //	                                           transformations  of the molecule are
-		}                             //	                                           contained in the datablock. Otherwise, blank.
+		} //	                                           contained in the datablock. Otherwise, blank.
 
 		// clang-format off
 		getCategory("struct_ncs_oper")->emplace({
@@ -6376,11 +6367,20 @@ void read_pdb_file(std::istream &pdbFile, cif::file &cifFile)
 
 	p.Parse(pdbFile, cifFile);
 
-	if (not cifFile.empty() and cifFile.front().get_validator() == nullptr)
-		cifFile.front().load_dictionary("mmcif_pdbx.dic");
+	if (cifFile.empty())
+	{
+		if (VERBOSE >= 0)
+			std::cerr << "PDB is empty!\n";
+	}
+	else
+	{
+		cifFile.front().load_dictionary();
+		if (cifFile.front().get_validator() == nullptr)
+			cifFile.front().set_validator(&validator_factory::instance().get("mmcif_pdbx.dic"));
 
-	if (not cifFile.is_valid() and cif::VERBOSE >= 0)
-		std::cerr << "Resulting mmCIF file is not valid!\n";
+		if (not cifFile.is_valid() and cif::VERBOSE >= 0)
+			std::cerr << "Resulting mmCIF file is not valid!\n";
+	}
 }
 
 // --------------------------------------------------------------------
@@ -6404,7 +6404,10 @@ file read(std::istream &is)
 		// apart from the letter 'd', the test has changed into the following:
 
 		if (std::isalpha(ch) and std::toupper(ch) != 'D')
+		{
 			read_pdb_file(is, result);
+			reconstruct_pdbx(result);
+		}
 		else
 		{
 			try
@@ -6415,16 +6418,32 @@ file read(std::istream &is)
 			{
 				std::throw_with_nested(std::runtime_error("Since the file did not start with a valid PDB HEADER line mmCIF was assumed, but that failed."));
 			}
-		}
 
-		// Since we're using the cif::pdb way of reading the file, the data may need
-		// reconstruction
-		reconstruct_pdbx(result);
+			if (not(result.empty() or result.front().empty()))
+			{
+				if (auto &db = result.front(); db.get("audit_conform") == nullptr)
+					reconstruct_pdbx(result);
+				else
+				{
+					try
+					{
+						// Try to see if we can create an mm::structure out of this data.
+						// If that fails, we need to reconstruct a PDBx file out of it.
+
+						cif::mm::structure s(result);
+					}
+					catch (const std::exception &e)
+					{
+						reconstruct_pdbx(result);
+					}
+				}
+			}
+		}
 	}
 
 	// Must be a PDB like file, right?
 	if (not result.empty() and result.front().get_validator() == nullptr)
-		result.front().load_dictionary("mmcif_pdbx.dic");
+		result.front().set_validator(&validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	return result;
 }

src/pdb/pdb2cif_remark_3.cpp

@@ -1478,6 +1478,8 @@ bool Remark3Parser::parse(const std::string &expMethod, PDBRecord *r, cif::datab
 
 		best.parser->fixup();
 
+		auto &validator = cif::validator_factory::instance().get("mmcif_pdbx.dic");
+
 		for (auto &cat1 : best.parser->mDb)
 		{
 			if (cat1.empty())
@@ -1496,8 +1498,15 @@ bool Remark3Parser::parse(const std::string &expMethod, PDBRecord *r, cif::datab
 					auto r1 = cat1.front();
 					auto r2 = cat2.front();
 
-					for (auto item : cat1.key_items())
-						r2[item] = r1[item].text();
+					auto cv = cat1.get_cat_validator();
+					if (cv == nullptr)
+						cv = validator.get_validator_for_category(cat1.name());
+					
+					if (cv == nullptr)
+						continue;
+
+					for (auto &iv : cv->m_item_validators)
+						r2[iv.m_item_name] = r1[iv.m_item_name].text();
 				}
 			}
 			else

src/pdb/reconstruct.cpp

@@ -104,7 +104,7 @@ void checkEntities(datablock &db)
 
 		float formula_weight = 0;
 
-		if (type.empty())	// yes, that happens
+		if (type.empty()) // yes, that happens
 		{
 			const auto comp_id = db["atom_site"].find_first<std::string>("label_entity_id"_key == entity_id, "label_comp_id");
 			auto compound = cf.create(comp_id);
@@ -125,10 +125,10 @@ void checkEntities(datablock &db)
 
 			if (type.empty())
 				throw std::runtime_error("Entity without type and cannot determine what it should be");
-			
+
 			entity["type"] = type;
 		}
-		
+
 		if (type == "polymer")
 		{
 			int n = 0;
@@ -136,10 +136,10 @@ void checkEntities(datablock &db)
 			for (std::string comp_id : db["pdbx_poly_seq_scheme"].find<std::string>("entity_id"_key == entity_id, "mon_id"))
 			{
 				auto compound = cf.create(comp_id);
-				assert(compound);
-				if (not compound)
-					throw std::runtime_error("missing information for compound " + comp_id);
-				formula_weight += compound->formula_weight();
+				if (compound)
+					formula_weight += compound->formula_weight();
+				else if (cif::VERBOSE > 0)
+					std::clog << "missing information for compound " + comp_id << '\n';
 				++n;
 			}
 
@@ -154,10 +154,10 @@ void checkEntities(datablock &db)
 			for (std::string comp_id : db["pdbx_entity_branch_list"].find<std::string>("entity_id"_key == entity_id, "comp_id"))
 			{
 				auto compound = cf.create(comp_id);
-				assert(compound);
-				if (not compound)
-					throw std::runtime_error("missing information for compound " + comp_id);
-				formula_weight += compound->formula_weight();
+				if (compound)
+					formula_weight += compound->formula_weight();
+				else if (cif::VERBOSE > 0)
+					std::clog << "missing information for compound " + comp_id << '\n';
 				++n;
 			}
 
@@ -205,11 +205,11 @@ void createEntityIDs(datablock &db)
 	std::vector<residue_key_type> waters;
 
 	for (residue_key_type k : atom_site.rows<std::optional<std::string>,
-							  std::optional<int>,
-							  std::optional<std::string>,
-							  std::optional<std::string>,
-							  std::optional<int>,
-							  std::optional<std::string>>(
+			 std::optional<int>,
+			 std::optional<std::string>,
+			 std::optional<std::string>,
+			 std::optional<int>,
+			 std::optional<std::string>>(
 			 "auth_asym_id", "auth_seq_id", "auth_comp_id",
 			 "label_asym_id", "label_seq_id", "label_comp_id"))
 	{
@@ -491,7 +491,7 @@ void checkAtomRecords(datablock &db)
 		auto chem_comp_entry = chem_comp.find_first("id"_key == comp_id);
 
 		std::optional<bool> non_std;
-		if  (cf.is_monomer(comp_id))
+		if (cf.is_monomer(comp_id))
 			non_std = cf.is_std_monomer(comp_id);
 
 		if (not chem_comp_entry)
@@ -776,6 +776,7 @@ void createEntity(datablock &db)
 void createEntityPoly(datablock &db)
 {
 	using namespace literals;
+	using namespace std::literals;
 
 	auto &cf = compound_factory::instance();
 
@@ -802,7 +803,7 @@ void createEntityPoly(datablock &db)
 			auto c = cf.create(comp_id);
 
 			std::string letter;
-			char letter_can;
+			char letter_can{};
 
 			// TODO: Perhaps we should improve this...
 			if (type != "other")
@@ -914,7 +915,7 @@ void createEntityPoly(datablock &db)
 
 		entity_poly.emplace({ //
 			{ "entity_id", entity_id },
-			{ "type", type },
+			{ "type", type.empty() ? "other"s : type },
 			{ "nstd_linkage", non_std_linkage },
 			{ "nstd_monomer", non_std_monomer },
 			{ "pdbx_seq_one_letter_code", entity_seq },
@@ -1190,7 +1191,71 @@ void createPdbxNonpolyScheme(datablock &db)
 	}
 }
 
-bool reconstruct_pdbx(file &file, std::string_view dictionary)
+void createPdbxBranchScheme(datablock &db)
+{
+	using namespace literals;
+
+	createPdbxEntityNonpoly(db);
+
+	auto &entity = db["entity"];
+	auto &pdbx_branch_scheme = db["pdbx_branch_scheme"];
+	auto &pdbx_entity_branch_list = db["pdbx_entity_branch_list"];
+	auto &atom_site = db["atom_site"];
+
+	for (const auto entity_id : entity.find<std::string>("type"_key == "branched", "id"))
+	{
+		for (const auto &[comp_id, asym_id, auth_seq_id] : atom_site.find<std::string, std::string, std::optional<int>>("label_entity_id"_key == entity_id, "label_comp_id", "label_asym_id", "auth_seq_id"))
+		{
+			if (not auth_seq_id.has_value())
+				throw std::runtime_error("Missing auth_seq_id on sugar atom");
+
+			int num = *auth_seq_id;
+
+			if (not pdbx_entity_branch_list.contains("entity_id"_key == entity_id and "num"_key == num))
+			{
+				pdbx_entity_branch_list.emplace({
+					// clang-format off
+
+					{ "entity_id", entity_id },
+					{ "comp_id", comp_id },
+					{ "num", num },
+
+					// clang-format on
+				});
+			}
+
+			if (not pdbx_branch_scheme.contains("entity_id"_key == entity_id and "asym_id"_key == asym_id and "num"_key == num))
+			{
+				pdbx_branch_scheme.emplace({
+					// clang-format off
+					{ "asym_id", asym_id },
+					{ "entity_id", entity_id },
+					{ "mon_id", comp_id },
+					{ "num", num },
+					{ "pdb_asym_id", asym_id },
+					{ "pdb_mon_id", comp_id },
+					{ "pdb_seq_num", num }
+					// clang-format on
+				});
+			}
+		}
+	}
+}
+
+bool reconstruct_pdbx(file &file)
+{
+	if (file.empty())
+		throw std::runtime_error("Cannot reconstruct PDBx, file seems to be empty");
+
+	auto &db = file.front();
+
+	if (auto ac = db.get("audit_conform"); ac != nullptr)
+		return reconstruct_pdbx(file, validator_factory::instance().get(*ac));
+	else
+		return reconstruct_pdbx(file, validator_factory::instance().get("mmcif_pdbx.dic"));
+}
+
+bool reconstruct_pdbx(file &file, const validator &validator)
 {
 	if (file.empty())
 		throw std::runtime_error("Cannot reconstruct PDBx, file seems to be empty");
@@ -1204,8 +1269,6 @@ bool reconstruct_pdbx(file &file, std::string_view dictionary)
 	if (auto cat = db.get("atom_site"); cat == nullptr or cat->empty())
 		throw std::runtime_error("Cannot reconstruct PDBx file, atom data missing");
 
-	auto &validator = validator_factory::instance()[dictionary];
-
 	std::string entry_id;
 
 	// Phenix files do not have an entry record
@@ -1456,7 +1519,7 @@ bool reconstruct_pdbx(file &file, std::string_view dictionary)
 
 	db["chem_comp"].reorder_by_index();
 
-	db.load_dictionary(dictionary);
+	db.set_validator(&validator);
 
 	if (db.get("atom_site_anisotrop"))
 		checkAtomAnisotropRecords(db);
@@ -1469,9 +1532,99 @@ bool reconstruct_pdbx(file &file, std::string_view dictionary)
 	if (auto cat = db.get("entity"); cat == nullptr or cat->empty())
 		createEntity(db);
 
+	if (auto cat = db.get("pdbx_poly_seq_scheme"); cat == nullptr or cat->empty())
+		createPdbxPolySeqScheme(db);
+
+	if (auto cat = db.get("ndb_poly_seq_scheme"); cat == nullptr or cat->empty())
+		comparePolySeqSchemes(db);
+	
+	createPdbxNonpolyScheme(db);
+
+	// Create a minimal set of branch records
+	createPdbxBranchScheme(db);
+
 	// fill in missing formula_weight, e.g.
 	checkEntities(db);
 
+	// skip unknown categories for now
+	bool valid = true;
+	for (auto &cat : db)
+		valid = valid and (cat.get_cat_validator() == nullptr or cat.is_valid());
+
+	return valid and is_valid_pdbx_file(file, validator);
+}
+
+// --------------------------------------------------------------------
+
+void fixup_pdbx(file &file)
+{
+	if (file.empty())
+		throw std::runtime_error("Cannot reconstruct PDBx, file seems to be empty");
+
+	auto &db = file.front();
+
+	if (auto ac = db.get("audit_conform"); ac != nullptr)
+		fixup_pdbx(file, validator_factory::instance().get(*ac));
+	else
+		fixup_pdbx(file, validator_factory::instance().get("mmcif_pdbx.dic"));
+}
+
+void fixup_pdbx(file &file, const validator &validator)
+{
+	if (file.empty())
+		throw std::runtime_error("Cannot reconstruct PDBx, file seems to be empty");
+
+	// assuming the first datablock contains the entry ...
+	auto &db = file.front();
+
+	// ... and any additional datablock will contain compound information
+	cif::compound_source cs(file);
+
+	if (auto cat = db.get("atom_site"); cat == nullptr or cat->empty())
+		throw std::runtime_error("Cannot reconstruct PDBx file, atom data missing");
+
+	std::string entry_id;
+
+	// Phenix files do not have an entry record
+	if (auto cat = db.get("entry"); cat == nullptr or cat->empty())
+	{
+		entry_id = db.name();
+		category entry("entry");
+		entry.emplace({ { "id", entry_id } });
+		db.emplace_back(std::move(entry));
+	}
+	else
+	{
+		auto &entry = db["entry"];
+		if (entry.size() != 1)
+			throw std::runtime_error("Unexpected size of entry category");
+
+		entry_id = entry.front().get<std::string>("id");
+	}
+
+	// Start with chem_comp, it is often missing many fields
+	// that can easily be filled in.
+	checkChemCompRecords(db);
+
+	// If the data is really horrible, it might not contain entities
+	if (not db["atom_site"].find_first(key("label_entity_id") != null))
+		createEntityIDs(db);
+
+	// Now see if atom records make sense at all
+	checkAtomRecords(db);
+
+	db["chem_comp"].reorder_by_index();
+
+	db.set_validator(&validator);
+
+	// Now create any missing categories
+	// Next make sure we have struct_asym records
+	if (auto cat = db.get("struct_asym"); cat == nullptr or cat->empty())
+		createStructAsym(db);
+
+	if (auto cat = db.get("entity"); cat == nullptr or cat->empty())
+		createEntity(db);
+
 	if (auto cat = db.get("pdbx_poly_seq_scheme"); cat == nullptr or cat->empty())
 		createPdbxPolySeqScheme(db);
 
@@ -1480,12 +1633,13 @@ bool reconstruct_pdbx(file &file, std::string_view dictionary)
 	
 	createPdbxNonpolyScheme(db);
 
-	// skip unknown categories for now
-	bool valid = true;
-	for (auto &cat : db)
-		valid = valid and (cat.get_cat_validator() == nullptr or cat.is_valid());
+	// Create a minimal set of branch records
+	createPdbxBranchScheme(db);
 
-	return valid and is_valid_pdbx_file(file, dictionary);
+	// fill in missing formula_weight, e.g.
+	checkEntities(db);
+
+	// That's it
 }
 
 } // namespace cif::pdb

src/pdb/validate-pdbx.cpp

@@ -67,10 +67,10 @@ condition get_parents_condition(const validator &validator, row_handle rh, const
 	return result;
 }
 
-bool is_valid_pdbx_file(const file &file, std::string_view dictionary)
+bool is_valid_pdbx_file(const file &file, const validator &v)
 {
 	std::error_code ec;
-	bool result = is_valid_pdbx_file(file, dictionary, ec);
+	bool result = is_valid_pdbx_file(file, v, ec);
 	return result and not (bool)ec;
 }
 
@@ -80,33 +80,15 @@ bool is_valid_pdbx_file(const file &file, std::error_code &ec)
 
 	if (file.empty())
 		ec = make_error_code(validation_error::empty_file);
+	else if (auto ac = file.front().get("audit_conform"); ac != nullptr)
+		result = is_valid_pdbx_file(file, validator_factory::instance().get(*ac), ec);
 	else
-	{
-		std::string dictionary = "mmcif_pdbx";
-
-		for (auto &db : file)
-		{
-			auto audit_conform = db.get("audit_conform");
-			if (audit_conform == nullptr)
-				continue;
-			
-			if (not audit_conform->empty())
-			{
-				auto specified_dict = audit_conform->front()["dict_name"];
-				if (not specified_dict.empty())
-					dictionary = specified_dict.as<std::string>();
-			}
-
-			break;
-		}
-
-		result = is_valid_pdbx_file(file, dictionary, ec);
-	}
+		result = is_valid_pdbx_file(file, validator_factory::instance().get("mmcif_pdbx.dic"), ec);
 	
 	return result;
 }
 
-bool is_valid_pdbx_file(const file &file, std::string_view dictionary, std::error_code &ec)
+bool is_valid_pdbx_file(const file &file, const validator &validator, std::error_code &ec)
 {
 	using namespace cif::literals;
 
@@ -115,7 +97,6 @@ bool is_valid_pdbx_file(const file &file, std::string_view dictionary, std::erro
 	try
 	{
 		auto &cf = cif::compound_factory::instance();
-		auto &validator = cif::validator_factory::instance().operator[](dictionary);
 
 		if (file.empty())
 			throw std::runtime_error("Empty file");
@@ -335,4 +316,4 @@ bool is_valid_pdbx_file(const file &file, std::string_view dictionary, std::erro
 }
 
 } // namespace cif::pdb
-  
+  

src/validate.cpp

@@ -25,6 +25,7 @@
  */
 
 #include "cif++/validate.hpp"
+#include "cif++/category.hpp"
 #include "cif++/dictionary_parser.hpp"
 #include "cif++/gzio.hpp"
 #include "cif++/utilities.hpp"
@@ -109,9 +110,15 @@ type_validator::type_validator(std::string_view name, DDL_PrimitiveType type, st
 {
 }
 
+type_validator::type_validator(const type_validator &tv)
+	: m_name(tv.m_name)
+	, m_primitive_type(tv.m_primitive_type)
+	, m_rx(tv.m_rx)
+{
+}
+
 type_validator::~type_validator()
 {
-	delete m_rx;
 }
 
 int type_validator::compare(std::string_view a, std::string_view b) const
@@ -242,7 +249,7 @@ void category_validator::add_item_validator(item_validator &&v)
 	if (v.m_mandatory)
 		m_mandatory_items.insert(v.m_item_name);
 
-	v.m_category = this;
+	v.m_category = m_name;
 
 	auto r = m_item_validators.insert(std::move(v));
 	if (not r.second and VERBOSE >= 4)
@@ -284,6 +291,29 @@ const item_validator *category_validator::get_validator_for_aliased_item(std::st
 
 // --------------------------------------------------------------------
 
+validator::validator(const validator &rhs)
+	: m_audit_conform(rhs.m_audit_conform)
+	, m_strict(rhs.m_strict)
+	, m_type_validators(rhs.m_type_validators)
+	, m_category_validators(rhs.m_category_validators)
+	, m_link_validators(rhs.m_link_validators)
+{
+}
+
+void swap(validator &a, validator &b) noexcept
+{
+	std::swap(a.m_audit_conform, b.m_audit_conform);
+	std::swap(a.m_strict, b.m_strict);
+	std::swap(a.m_type_validators, b.m_type_validators);
+	std::swap(a.m_category_validators, b.m_category_validators);
+	std::swap(a.m_link_validators, b.m_link_validators);
+}
+
+void validator::parse(std::istream &is)
+{
+	parse_dictionary(*this, is);
+}
+
 void validator::add_type_validator(type_validator &&v)
 {
 	auto r = m_type_validators.insert(std::move(v));
@@ -416,6 +446,48 @@ void validator::report_error(std::error_code ec, std::string_view category,
 		std::cerr << ex.what() << '\n';
 }
 
+void validator::fill_audit_conform(category &audit_conform) const
+{
+	audit_conform.clear();
+	audit_conform.emplace(m_audit_conform.begin(), m_audit_conform.end());
+}
+
+bool validator::matches_audit_conform(const category &audit_conform) const
+{
+	if (audit_conform.empty())
+		return false;
+
+	auto ai = m_audit_conform.begin();
+	auto bi = audit_conform.begin();
+
+	while (ai != m_audit_conform.end() and bi != audit_conform.end())
+	{
+		const auto &[name_a, version_a] = ai->get<std::string, std::optional<std::string>>("dict_name", "dict_version");
+		const auto &[name_b, version_b] = bi->get<std::string, std::optional<std::string>>("dict_name", "dict_version");
+
+		++ai;
+		++bi;
+
+		if (name_a != name_b)
+			return false;
+
+		if (not version_b.has_value() or not version_a.has_value())
+			continue;
+
+		if (validator_factory::check_version(name_a, *version_b, *version_a) == false)
+			return false;
+	}
+
+	return ai == m_audit_conform.end() and bi == audit_conform.end();
+}
+
+void validator::append_audit_conform(const std::string &name, const std::optional<std::string> &version)
+{
+	m_audit_conform.emplace({ //
+		{ "dict_name", name },
+		{ "dict_version", version } });
+}
+
 // --------------------------------------------------------------------
 
 validator_factory &validator_factory::instance()
@@ -424,113 +496,122 @@ validator_factory &validator_factory::instance()
 	return s_instance;
 }
 
-const validator &validator_factory::operator[](std::string_view dictionary_name)
+const validator &validator_factory::get(std::string_view dictionary_name)
 {
-	try
-	{
-		std::lock_guard lock(m_mutex);
+	category audit_conform("audit_conform");
+	for (auto part : cif::split(dictionary_name, ";", true))
+		audit_conform.emplace({ { "dict_name", part } });
 
-		for (auto &validator : m_validators)
-		{
-			if (iequals(validator.name(), dictionary_name))
-				return validator;
-		}
-
-		// not found, try to see if it helps if we tweak the name a little
-
-		// too bad clang version 10 did not have a constructor for std::filesystem::path that accepts a std::string_view
-		std::filesystem::path dictionary(dictionary_name.data(), dictionary_name.data() + dictionary_name.length());
-
-		if (dictionary.extension() != ".dic")
-		{
-			auto dict_name = dictionary.filename().string() + ".dic";
-
-			for (auto &validator : m_validators)
-			{
-				if (iequals(validator.name(), dict_name))
-					return validator;
-			}
-		}
-
-		// not found, add it
-
-		validator v(dictionary_name);
-
-		for (bool first = true; auto part_name : cif::split(dictionary_name, ";", true))
-		{
-			auto data = load_resource(part_name);
-			dictionary.assign(part_name.begin(), part_name.end());
-
-			if (not data and dictionary.extension().string() != ".dic")
-				data = load_resource(dictionary.parent_path() / (dictionary.filename().string() + ".dic"));
-
-			if (not data)
-			{
-				std::error_code ec;
-
-				// might be a compressed dictionary on disk
-				std::filesystem::path p = dictionary;
-				if (p.extension() == ".dic")
-					p = p.parent_path() / (p.filename().string() + ".gz");
-				else
-					p = p.parent_path() / (p.filename().string() + ".dic.gz");
-
-#if defined(CACHE_DIR) or defined(DATA_DIR)
-				if (not std::filesystem::exists(p, ec) or ec)
-				{
-					for (const char *dir : {
-# if defined(CACHE_DIR)
-							 CACHE_DIR,
-# endif
-# if defined(DATA_DIR)
-							 DATA_DIR
-# endif
-						 })
-					{
-						auto p2 = std::filesystem::path(dir) / p;
-						if (std::filesystem::exists(p2, ec) and not ec)
-						{
-							swap(p, p2);
-							break;
-						}
-					}
-				}
-#endif
-
-				if (std::filesystem::exists(p, ec) and not ec)
-				{
-					auto in = std::make_unique<gzio::ifstream>(p);
-
-					if (not in->is_open())
-						throw std::runtime_error("Could not open dictionary (" + p.string() + ")");
-
-					data.reset(in.release());
-				}
-				else
-					throw std::runtime_error("Dictionary not found or defined (" + dictionary.string() + ")");
-			}
-
-			if (std::exchange(first, false))
-				v = parse_dictionary(part_name, *data);
-			else
-				extend_dictionary(v, *data);
-		}
-
-		m_validators.emplace_back(std::move(v));
-
-		return m_validators.back();
-	}
-	catch (const std::exception &ex)
-	{
-		std::string msg = "Error while loading dictionary ";
-		msg += dictionary_name;
-		std::throw_with_nested(std::runtime_error(msg));
-	}
+	return get(audit_conform);
 }
 
-const validator &validator_factory::construct_validator(std::string_view name, std::istream &is)
+const validator &validator_factory::get(const category &audit_conform)
 {
-	return m_validators.emplace_back(parse_dictionary(name, is));
+	if (audit_conform.empty())
+		throw std::runtime_error("Empty audit_conform category, cannot create a validator");
+
+	std::lock_guard lock(m_mutex);
+
+	// Check existing first
+	for (auto &v : m_validators)
+	{
+		if (v.matches_audit_conform(audit_conform))
+			return v;
+	}
+
+	// If the audit conform contains only one record, this is easy
+	if (audit_conform.size() == 1)
+	{
+		const auto &[name, version] = audit_conform.front().get<std::string, std::optional<std::string>>("dict_name", "dict_version");
+		return m_validators.emplace_back(construct_validator(name, version));
+	}
+
+	// A new, merged dictionary
+
+	std::optional<validator> v;
+	for (const auto &[name, version] : audit_conform.rows<std::string, std::optional<std::string>>("dict_name", "dict_version"))
+	{
+		if (not v)
+			v = construct_validator(name, version);
+		else
+		{
+			auto data = load_resource(name);
+			if (not data)
+				throw std::runtime_error("Could not load dictionary " + std::string{ name });
+
+			v->parse(*data);
+		}
+	}
+
+	if (not v)
+		throw std::runtime_error("Missing dictionary information?");
+
+	return m_validators.emplace_back(std::move(*v));
+}
+
+validator validator_factory::construct_validator(std::string_view name, std::optional<std::string> version)
+{
+	auto data = load_resource(name);
+	if (not data and name == "mmcif_pdbx_v50")
+		data = load_resource("mmcif_pdbx.dic");
+
+	if (not data)
+		throw std::runtime_error("Could not load dictionary " + std::string{ name });
+
+	validator v;
+	v.parse(*data);
+
+	if (version.has_value() and VERBOSE >= 0 and
+		not v.matches_audit_conform(category{ "audit_conform", //
+			{ { "dict_name", name }, { "dict_version", version } } }))
+	{
+		std::clog << "Invalid dictionary?\n";
+	}
+
+	return v;
+}
+
+bool validator_factory::check_version(std::string_view name, std::string_view expected, std::string_view found)
+{
+	bool result = true;
+	auto el = cif::split(expected, ".");
+	auto fl = cif::split(found, ".");
+
+	auto eli = el.begin();
+	auto fli = fl.begin();
+
+	while (eli != el.end() and fli != fl.end())
+	{
+		int e_int, f_int;
+		if (auto [ptr, ec] = std::from_chars(eli->data(), eli->data() + eli->length(), e_int); ec != std::errc{})
+		{
+			std::clog << "Could not parse requested version string for dictionary " << std::quoted(expected) << "\n";
+			result = false;
+			break;
+		}
+
+		if (auto [ptr, ec] = std::from_chars(fli->data(), fli->data() + fli->length(), f_int); ec != std::errc{})
+		{
+			std::clog << "Could not parse version string in dictionary " << name << " " << std::quoted(found) << "\n";
+			result = false;
+			break;
+		}
+
+		if (f_int > e_int) // newer version, assume this is ok
+			break;
+
+		if (f_int < e_int)
+		{
+			std::clog << "The version in dictionary " << name << " is lower than requested, this may cause validation errors\n";
+			result = false;
+			break;
+		}
+
+		++eli;
+		++fli;
+	}
+
+	return result;
 }
 
 } // namespace cif

test/CMakeLists.txt

@@ -28,7 +28,8 @@ list(
 	sugar
 	spinner
 	# reconstruction
-	validate-pdbx)
+	validate-pdbx
+	)
 
 add_library(test-main OBJECT "${CMAKE_CURRENT_SOURCE_DIR}/test-main.cpp")
 

test/brak.pdb

@@ -0,0 +1,822 @@
+REMARK 001 design 0 N 3 RMSD 1.605
+MODEL       1
+ATOM      1  N   MET A   1     -24.216  -7.571  -1.155  1.00 31.87           N
+ATOM      2  CA  MET A   1     -22.971  -7.756  -0.416  1.00 31.87           C
+ATOM      3  C   MET A   1     -21.814  -7.051  -1.114  1.00 31.87           C
+ATOM      4  CB  MET A   1     -22.656  -9.245  -0.256  1.00 31.87           C
+ATOM      5  O   MET A   1     -21.516  -7.338  -2.275  1.00 31.87           O
+ATOM      6  CG  MET A   1     -22.779  -9.747   1.174  1.00 31.87           C
+ATOM      7  SD  MET A   1     -22.473 -11.551   1.314  1.00 31.87           S
+ATOM      8  CE  MET A   1     -22.597 -11.758   3.112  1.00 31.87           C
+ATOM      9  N   LYS A   2     -21.652  -5.596  -0.909  1.00 33.06           N
+ATOM     10  CA  LYS A   2     -21.035  -4.320  -0.554  1.00 33.06           C
+ATOM     11  C   LYS A   2     -19.531  -4.475  -0.351  1.00 33.06           C
+ATOM     12  CB  LYS A   2     -21.678  -3.747   0.710  1.00 33.06           C
+ATOM     13  O   LYS A   2     -19.087  -5.326   0.423  1.00 33.06           O
+ATOM     14  CG  LYS A   2     -22.957  -2.963   0.453  1.00 33.06           C
+ATOM     15  CD  LYS A   2     -23.454  -2.275   1.717  1.00 33.06           C
+ATOM     16  CE  LYS A   2     -24.810  -1.615   1.500  1.00 33.06           C
+ATOM     17  NZ  LYS A   2     -25.284  -0.906   2.725  1.00 33.06           N
+ATOM     18  N   ILE A   3     -18.889  -4.065  -1.586  1.00 34.22           N
+ATOM     19  CA  ILE A   3     -17.539  -3.550  -1.381  1.00 34.22           C
+ATOM     20  C   ILE A   3     -17.331  -2.297  -2.229  1.00 34.22           C
+ATOM     21  CB  ILE A   3     -16.471  -4.612  -1.721  1.00 34.22           C
+ATOM     22  O   ILE A   3     -17.508  -2.330  -3.449  1.00 34.22           O
+ATOM     23  CG1 ILE A   3     -16.524  -5.766  -0.713  1.00 34.22           C
+ATOM     24  CG2 ILE A   3     -15.075  -3.982  -1.759  1.00 34.22           C
+ATOM     25  CD1 ILE A   3     -15.743  -7.000  -1.142  1.00 34.22           C
+ATOM     26  N   ILE A   4     -17.776  -0.985  -1.831  1.00 40.18           N
+ATOM     27  CA  ILE A   4     -17.255   0.212  -1.181  1.00 40.18           C
+ATOM     28  C   ILE A   4     -16.631   1.136  -2.224  1.00 40.18           C
+ATOM     29  CB  ILE A   4     -16.218  -0.142  -0.091  1.00 40.18           C
+ATOM     30  O   ILE A   4     -15.717   0.736  -2.949  1.00 40.18           O
+ATOM     31  CG1 ILE A   4     -16.844  -1.063   0.963  1.00 40.18           C
+ATOM     32  CG2 ILE A   4     -15.658   1.128   0.555  1.00 40.18           C
+ATOM     33  CD1 ILE A   4     -15.839  -1.662   1.937  1.00 40.18           C
+ATOM     34  N   GLU A   5     -17.325   2.198  -2.769  1.00 38.95           N
+ATOM     35  CA  GLU A   5     -17.243   3.640  -2.553  1.00 38.95           C
+ATOM     36  C   GLU A   5     -15.944   4.019  -1.848  1.00 38.95           C
+ATOM     37  CB  GLU A   5     -18.445   4.131  -1.741  1.00 38.95           C
+ATOM     38  O   GLU A   5     -15.602   3.445  -0.812  1.00 38.95           O
+ATOM     39  CG  GLU A   5     -19.296   5.162  -2.466  1.00 38.95           C
+ATOM     40  CD  GLU A   5     -20.407   5.738  -1.602  1.00 38.95           C
+ATOM     41  OE1 GLU A   5     -20.194   6.791  -0.960  1.00 38.95           O
+ATOM     42  OE2 GLU A   5     -21.500   5.129  -1.566  1.00 38.95           O
+ATOM     43  N   LYS A   6     -14.697   4.048  -2.514  1.00 40.24           N
+ATOM     44  CA  LYS A   6     -13.793   5.173  -2.294  1.00 40.24           C
+ATOM     45  C   LYS A   6     -12.348   4.784  -2.593  1.00 40.24           C
+ATOM     46  CB  LYS A   6     -13.911   5.684  -0.857  1.00 40.24           C
+ATOM     47  O   LYS A   6     -11.769   3.946  -1.899  1.00 40.24           O
+ATOM     48  CG  LYS A   6     -14.799   6.911  -0.707  1.00 40.24           C
+ATOM     49  CD  LYS A   6     -14.696   7.508   0.690  1.00 40.24           C
+ATOM     50  CE  LYS A   6     -15.689   8.645   0.887  1.00 40.24           C
+ATOM     51  NZ  LYS A   6     -15.523   9.301   2.219  1.00 40.24           N
+ATOM     52  N   TYR A   7     -12.096   4.360  -3.877  1.00 55.19           N
+ATOM     53  CA  TYR A   7     -10.695   4.330  -4.282  1.00 55.19           C
+ATOM     54  C   TYR A   7      -9.919   5.478  -3.647  1.00 55.19           C
+ATOM     55  CB  TYR A   7     -10.574   4.398  -5.807  1.00 55.19           C
+ATOM     56  O   TYR A   7     -10.324   6.639  -3.746  1.00 55.19           O
+ATOM     57  CG  TYR A   7     -11.069   3.159  -6.511  1.00 55.19           C
+ATOM     58  CD1 TYR A   7     -10.245   2.047  -6.669  1.00 55.19           C
+ATOM     59  CD2 TYR A   7     -12.361   3.097  -7.021  1.00 55.19           C
+ATOM     60  CE1 TYR A   7     -10.696   0.903  -7.319  1.00 55.19           C
+ATOM     61  CE2 TYR A   7     -12.823   1.958  -7.673  1.00 55.19           C
+ATOM     62  OH  TYR A   7     -12.437  -0.262  -8.461  1.00 55.19           O
+ATOM     63  CZ  TYR A   7     -11.985   0.868  -7.817  1.00 55.19           C
+ATOM     64  N   VAL A   8      -9.308   5.253  -2.484  1.00 55.50           N
+ATOM     65  CA  VAL A   8      -8.469   6.226  -1.792  1.00 55.50           C
+ATOM     66  C   VAL A   8      -7.196   6.476  -2.597  1.00 55.50           C
+ATOM     67  CB  VAL A   8      -8.112   5.754  -0.364  1.00 55.50           C
+ATOM     68  O   VAL A   8      -6.488   5.534  -2.960  1.00 55.50           O
+ATOM     69  CG1 VAL A   8      -7.215   6.774   0.333  1.00 55.50           C
+ATOM     70  CG2 VAL A   8      -9.381   5.509   0.450  1.00 55.50           C
+ATOM     71  N   PHE A   9      -7.225   7.625  -3.250  1.00 66.91           N
+ATOM     72  CA  PHE A   9      -6.004   8.094  -3.895  1.00 66.91           C
+ATOM     73  C   PHE A   9      -4.934   8.414  -2.858  1.00 66.91           C
+ATOM     74  CB  PHE A   9      -6.289   9.330  -4.754  1.00 66.91           C
+ATOM     75  O   PHE A   9      -5.089   9.345  -2.065  1.00 66.91           O
+ATOM     76  CG  PHE A   9      -6.868   9.009  -6.105  1.00 66.91           C
+ATOM     77  CD1 PHE A   9      -6.039   8.797  -7.200  1.00 66.91           C
+ATOM     78  CD2 PHE A   9      -8.243   8.920  -6.281  1.00 66.91           C
+ATOM     79  CE1 PHE A   9      -6.572   8.499  -8.452  1.00 66.91           C
+ATOM     80  CE2 PHE A   9      -8.783   8.623  -7.530  1.00 66.91           C
+ATOM     81  CZ  PHE A   9      -7.946   8.414  -8.614  1.00 66.91           C
+ATOM     82  N   LEU A  10      -4.065   7.388  -2.607  1.00 72.14           N
+ATOM     83  CA  LEU A  10      -2.967   7.542  -1.658  1.00 72.14           C
+ATOM     84  C   LEU A  10      -2.342   8.929  -1.771  1.00 72.14           C
+ATOM     85  CB  LEU A  10      -1.902   6.469  -1.893  1.00 72.14           C
+ATOM     86  O   LEU A  10      -1.903   9.499  -0.770  1.00 72.14           O
+ATOM     87  CG  LEU A  10      -2.222   5.069  -1.365  1.00 72.14           C
+ATOM     88  CD1 LEU A  10      -1.258   4.046  -1.957  1.00 72.14           C
+ATOM     89  CD2 LEU A  10      -2.165   5.045   0.158  1.00 72.14           C
+ATOM     90  N   ALA A  11      -2.498   9.659  -2.947  1.00 69.94           N
+ATOM     91  CA  ALA A  11      -1.847  10.939  -3.211  1.00 69.94           C
+ATOM     92  C   ALA A  11      -2.571  12.080  -2.501  1.00 69.94           C
+ATOM     93  CB  ALA A  11      -1.786  11.206  -4.713  1.00 69.94           C
+ATOM     94  O   ALA A  11      -1.971  13.117  -2.207  1.00 69.94           O
+ATOM     95  N   GLU A  12      -3.713  11.866  -1.974  1.00 76.23           N
+ATOM     96  CA  GLU A  12      -4.520  12.921  -1.368  1.00 76.23           C
+ATOM     97  C   GLU A  12      -4.620  12.741   0.144  1.00 76.23           C
+ATOM     98  CB  GLU A  12      -5.920  12.951  -1.987  1.00 76.23           C
+ATOM     99  O   GLU A  12      -5.178  13.590   0.841  1.00 76.23           O
+ATOM    100  CG  GLU A  12      -5.939  13.391  -3.444  1.00 76.23           C
+ATOM    101  CD  GLU A  12      -7.343  13.548  -4.005  1.00 76.23           C
+ATOM    102  OE1 GLU A  12      -7.904  14.665  -3.935  1.00 76.23           O
+ATOM    103  OE2 GLU A  12      -7.888  12.544  -4.518  1.00 76.23           O
+ATOM    104  N   LEU A  13      -3.986  11.682   0.620  1.00 82.70           N
+ATOM    105  CA  LEU A  13      -4.106  11.386   2.043  1.00 82.70           C
+ATOM    106  C   LEU A  13      -2.951  12.001   2.826  1.00 82.70           C
+ATOM    107  CB  LEU A  13      -4.146   9.873   2.274  1.00 82.70           C
+ATOM    108  O   LEU A  13      -1.826  12.073   2.327  1.00 82.70           O
+ATOM    109  CG  LEU A  13      -5.347   9.129   1.688  1.00 82.70           C
+ATOM    110  CD1 LEU A  13      -5.171   7.623   1.857  1.00 82.70           C
+ATOM    111  CD2 LEU A  13      -6.640   9.598   2.346  1.00 82.70           C
+ATOM    112  N   SER A  14      -3.339  12.631   3.947  1.00 86.03           N
+ATOM    113  CA  SER A  14      -2.328  13.121   4.878  1.00 86.03           C
+ATOM    114  C   SER A  14      -1.680  11.974   5.645  1.00 86.03           C
+ATOM    115  CB  SER A  14      -2.942  14.120   5.859  1.00 86.03           C
+ATOM    116  O   SER A  14      -2.161  10.839   5.599  1.00 86.03           O
+ATOM    117  OG  SER A  14      -3.526  15.211   5.168  1.00 86.03           O
+ATOM    118  N   ASP A  15      -0.488  12.215   6.359  1.00 86.85           N
+ATOM    119  CA  ASP A  15       0.221  11.246   7.189  1.00 86.85           C
+ATOM    120  C   ASP A  15      -0.734  10.542   8.149  1.00 86.85           C
+ATOM    121  CB  ASP A  15       1.345  11.930   7.971  1.00 86.85           C
+ATOM    122  O   ASP A  15      -0.647   9.327   8.340  1.00 86.85           O
+ATOM    123  CG  ASP A  15       2.418  12.521   7.074  1.00 86.85           C
+ATOM    124  OD1 ASP A  15       3.279  11.766   6.575  1.00 86.85           O
+ATOM    125  OD2 ASP A  15       2.403  13.754   6.867  1.00 86.85           O
+ATOM    126  N   GLU A  16      -1.606  11.353   8.835  1.00 86.30           N
+ATOM    127  CA  GLU A  16      -2.590  10.797   9.760  1.00 86.30           C
+ATOM    128  C   GLU A  16      -3.555   9.859   9.041  1.00 86.30           C
+ATOM    129  CB  GLU A  16      -3.366  11.918  10.456  1.00 86.30           C
+ATOM    130  O   GLU A  16      -3.871   8.780   9.545  1.00 86.30           O
+ATOM    131  CG  GLU A  16      -2.547  12.688  11.481  1.00 86.30           C
+ATOM    132  CD  GLU A  16      -3.342  13.770  12.194  1.00 86.30           C
+ATOM    133  OE1 GLU A  16      -3.925  13.489  13.266  1.00 86.30           O
+ATOM    134  OE2 GLU A  16      -3.384  14.908  11.676  1.00 86.30           O
+ATOM    135  N   GLU A  17      -4.022  10.266   7.880  1.00 87.05           N
+ATOM    136  CA  GLU A  17      -4.956   9.466   7.093  1.00 87.05           C
+ATOM    137  C   GLU A  17      -4.305   8.172   6.613  1.00 87.05           C
+ATOM    138  CB  GLU A  17      -5.477  10.268   5.898  1.00 87.05           C
+ATOM    139  O   GLU A  17      -4.937   7.114   6.618  1.00 87.05           O
+ATOM    140  CG  GLU A  17      -6.410  11.407   6.281  1.00 87.05           C
+ATOM    141  CD  GLU A  17      -7.322  11.844   5.146  1.00 87.05           C
+ATOM    142  OE1 GLU A  17      -8.505  11.433   5.125  1.00 87.05           O
+ATOM    143  OE2 GLU A  17      -6.850  12.602   4.270  1.00 87.05           O
+ATOM    144  N   LEU A  18      -3.057   8.282   6.290  1.00 88.16           N
+ATOM    145  CA  LEU A  18      -2.291   7.120   5.855  1.00 88.16           C
+ATOM    146  C   LEU A  18      -2.194   6.084   6.970  1.00 88.16           C
+ATOM    147  CB  LEU A  18      -0.889   7.539   5.407  1.00 88.16           C
+ATOM    148  O   LEU A  18      -2.374   4.888   6.730  1.00 88.16           O
+ATOM    149  CG  LEU A  18      -0.785   8.201   4.032  1.00 88.16           C
+ATOM    150  CD1 LEU A  18       0.627   8.730   3.803  1.00 88.16           C
+ATOM    151  CD2 LEU A  18      -1.179   7.219   2.934  1.00 88.16           C
+ATOM    152  N   LYS A  19      -1.979   6.558   8.141  1.00 87.30           N
+ATOM    153  CA  LYS A  19      -1.908   5.682   9.308  1.00 87.30           C
+ATOM    154  C   LYS A  19      -3.243   4.984   9.553  1.00 87.30           C
+ATOM    155  CB  LYS A  19      -1.495   6.474  10.549  1.00 87.30           C
+ATOM    156  O   LYS A  19      -3.280   3.787   9.843  1.00 87.30           O
+ATOM    157  CG  LYS A  19      -0.007   6.781  10.620  1.00 87.30           C
+ATOM    158  CD  LYS A  19       0.352   7.515  11.906  1.00 87.30           C
+ATOM    159  CE  LYS A  19       1.818   7.927  11.926  1.00 87.30           C
+ATOM    160  NZ  LYS A  19       2.176   8.639  13.189  1.00 87.30           N
+ATOM    161  N   LYS A  20      -4.246   5.781   9.525  1.00 88.15           N
+ATOM    162  CA  LYS A  20      -5.586   5.234   9.715  1.00 88.15           C
+ATOM    163  C   LYS A  20      -5.907   4.185   8.654  1.00 88.15           C
+ATOM    164  CB  LYS A  20      -6.631   6.350   9.682  1.00 88.15           C
+ATOM    165  O   LYS A  20      -6.511   3.153   8.956  1.00 88.15           O
+ATOM    166  CG  LYS A  20      -6.661   7.208  10.937  1.00 88.15           C
+ATOM    167  CD  LYS A  20      -7.772   8.249  10.877  1.00 88.15           C
+ATOM    168  CE  LYS A  20      -7.774   9.138  12.113  1.00 88.15           C
+ATOM    169  NZ  LYS A  20      -8.875  10.146  12.069  1.00 88.15           N
+ATOM    170  N   LEU A  21      -5.444   4.518   7.462  1.00 86.41           N
+ATOM    171  CA  LEU A  21      -5.662   3.611   6.341  1.00 86.41           C
+ATOM    172  C   LEU A  21      -4.974   2.271   6.583  1.00 86.41           C
+ATOM    173  CB  LEU A  21      -5.148   4.234   5.040  1.00 86.41           C
+ATOM    174  O   LEU A  21      -5.577   1.214   6.385  1.00 86.41           O
+ATOM    175  CG  LEU A  21      -5.364   3.417   3.765  1.00 86.41           C
+ATOM    176  CD1 LEU A  21      -6.842   3.400   3.391  1.00 86.41           C
+ATOM    177  CD2 LEU A  21      -4.524   3.976   2.622  1.00 86.41           C
+ATOM    178  N   VAL A  22      -3.737   2.304   6.914  1.00 87.19           N
+ATOM    179  CA  VAL A  22      -2.953   1.109   7.205  1.00 87.19           C
+ATOM    180  C   VAL A  22      -3.627   0.307   8.316  1.00 87.19           C
+ATOM    181  CB  VAL A  22      -1.504   1.464   7.609  1.00 87.19           C
+ATOM    182  O   VAL A  22      -3.719  -0.921   8.236  1.00 87.19           O
+ATOM    183  CG1 VAL A  22      -0.752   0.218   8.073  1.00 87.19           C
+ATOM    184  CG2 VAL A  22      -0.774   2.130   6.444  1.00 87.19           C
+ATOM    185  N   GLU A  23      -4.070   0.928   9.305  1.00 88.34           N
+ATOM    186  CA  GLU A  23      -4.790   0.285  10.400  1.00 88.34           C
+ATOM    187  C   GLU A  23      -6.050  -0.414   9.897  1.00 88.34           C
+ATOM    188  CB  GLU A  23      -5.151   1.308  11.480  1.00 88.34           C
+ATOM    189  O   GLU A  23      -6.364  -1.525  10.328  1.00 88.34           O
+ATOM    190  CG  GLU A  23      -3.959   1.792  12.292  1.00 88.34           C
+ATOM    191  CD  GLU A  23      -4.356   2.595  13.521  1.00 88.34           C
+ATOM    192  OE1 GLU A  23      -4.486   2.004  14.616  1.00 88.34           O
+ATOM    193  OE2 GLU A  23      -4.539   3.826  13.386  1.00 88.34           O
+ATOM    194  N   GLU A  24      -6.671   0.256   9.129  1.00 87.82           N
+ATOM    195  CA  GLU A  24      -7.883  -0.315   8.549  1.00 87.82           C
+ATOM    196  C   GLU A  24      -7.569  -1.568   7.737  1.00 87.82           C
+ATOM    197  CB  GLU A  24      -8.595   0.717   7.670  1.00 87.82           C
+ATOM    198  O   GLU A  24      -8.334  -2.535   7.758  1.00 87.82           O
+ATOM    199  CG  GLU A  24      -9.372   1.763   8.456  1.00 87.82           C
+ATOM    200  CD  GLU A  24     -10.598   1.200   9.157  1.00 87.82           C
+ATOM    201  OE1 GLU A  24     -10.544   0.974  10.387  1.00 87.82           O
+ATOM    202  OE2 GLU A  24     -11.621   0.982   8.469  1.00 87.82           O
+ATOM    203  N   TRP A  25      -6.452  -1.421   6.925  1.00 87.98           N
+ATOM    204  CA  TRP A  25      -6.008  -2.576   6.152  1.00 87.98           C
+ATOM    205  C   TRP A  25      -5.804  -3.789   7.054  1.00 87.98           C
+ATOM    206  CB  TRP A  25      -4.712  -2.256   5.403  1.00 87.98           C
+ATOM    207  O   TRP A  25      -6.237  -4.895   6.724  1.00 87.98           O
+ATOM    208  CG  TRP A  25      -4.867  -1.219   4.332  1.00 87.98           C
+ATOM    209  CD1 TRP A  25      -6.030  -0.819   3.734  1.00 87.98           C
+ATOM    210  CD2 TRP A  25      -3.820  -0.453   3.727  1.00 87.98           C
+ATOM    211  CE2 TRP A  25      -4.422   0.394   2.770  1.00 87.98           C
+ATOM    212  CE3 TRP A  25      -2.431  -0.399   3.902  1.00 87.98           C
+ATOM    213  NE1 TRP A  25      -5.769   0.151   2.794  1.00 87.98           N
+ATOM    214  CH2 TRP A  25      -2.324   1.317   2.182  1.00 87.98           C
+ATOM    215  CZ2 TRP A  25      -3.681   1.285   1.990  1.00 87.98           C
+ATOM    216  CZ3 TRP A  25      -1.695   0.488   3.125  1.00 87.98           C
+ATOM    217  N   ILE A  26      -5.164  -3.521   8.121  1.00 86.80           N
+ATOM    218  CA  ILE A  26      -4.842  -4.578   9.074  1.00 86.80           C
+ATOM    219  C   ILE A  26      -6.123  -5.095   9.724  1.00 86.80           C
+ATOM    220  CB  ILE A  26      -3.855  -4.082  10.155  1.00 86.80           C
+ATOM    221  O   ILE A  26      -6.287  -6.302   9.913  1.00 86.80           O
+ATOM    222  CG1 ILE A  26      -2.511  -3.706   9.519  1.00 86.80           C
+ATOM    223  CG2 ILE A  26      -3.667  -5.144  11.243  1.00 86.80           C
+ATOM    224  CD1 ILE A  26      -1.549  -3.010  10.471  1.00 86.80           C
+ATOM    225  N   LYS A  27      -7.085  -4.130  10.074  1.00 88.23           N
+ATOM    226  CA  LYS A  27      -8.358  -4.482  10.697  1.00 88.23           C
+ATOM    227  C   LYS A  27      -9.231  -5.292   9.743  1.00 88.23           C
+ATOM    228  CB  LYS A  27      -9.101  -3.224  11.147  1.00 88.23           C
+ATOM    229  O   LYS A  27      -9.857  -6.273  10.148  1.00 88.23           O
+ATOM    230  CG  LYS A  27     -10.280  -3.498  12.070  1.00 88.23           C
+ATOM    231  CD  LYS A  27     -10.890  -2.206  12.597  1.00 88.23           C
+ATOM    232  CE  LYS A  27     -12.096  -2.477  13.486  1.00 88.23           C
+ATOM    233  NZ  LYS A  27     -12.726  -1.212  13.967  1.00 88.23           N
+ATOM    234  N   SER A  28      -9.257  -4.866   8.486  1.00 88.60           N
+ATOM    235  CA  SER A  28     -10.097  -5.462   7.453  1.00 88.60           C
+ATOM    236  C   SER A  28      -9.499  -6.767   6.939  1.00 88.60           C
+ATOM    237  CB  SER A  28     -10.292  -4.488   6.290  1.00 88.60           C
+ATOM    238  O   SER A  28     -10.208  -7.598   6.366  1.00 88.60           O
+ATOM    239  OG  SER A  28     -11.021  -3.347   6.707  1.00 88.60           O
+ATOM    240  N   LYS A  29      -8.274  -7.060   7.266  1.00 87.22           N
+ATOM    241  CA  LYS A  29      -7.552  -8.279   6.911  1.00 87.22           C
+ATOM    242  C   LYS A  29      -7.546  -8.493   5.400  1.00 87.22           C
+ATOM    243  CB  LYS A  29      -8.169  -9.491   7.609  1.00 87.22           C
+ATOM    244  O   LYS A  29      -7.336  -9.612   4.927  1.00 87.22           O
+ATOM    245  CG  LYS A  29      -7.958  -9.511   9.116  1.00 87.22           C
+ATOM    246  CD  LYS A  29      -8.508 -10.787   9.741  1.00 87.22           C
+ATOM    247  CE  LYS A  29      -8.322 -10.795  11.252  1.00 87.22           C
+ATOM    248  NZ  LYS A  29      -8.836 -12.056  11.866  1.00 87.22           N
+ATOM    249  N   GLU A  30      -7.995  -7.561   4.711  1.00 90.12           N
+ATOM    250  CA  GLU A  30      -7.955  -7.620   3.253  1.00 90.12           C
+ATOM    251  C   GLU A  30      -7.715  -6.239   2.650  1.00 90.12           C
+ATOM    252  CB  GLU A  30      -9.254  -8.215   2.703  1.00 90.12           C
+ATOM    253  O   GLU A  30      -8.286  -5.248   3.110  1.00 90.12           O
+ATOM    254  CG  GLU A  30      -9.273  -8.357   1.188  1.00 90.12           C
+ATOM    255  CD  GLU A  30     -10.542  -9.008   0.662  1.00 90.12           C
+ATOM    256  OE1 GLU A  30     -11.585  -8.321   0.575  1.00 90.12           O
+ATOM    257  OE2 GLU A  30     -10.494 -10.215   0.335  1.00 90.12           O
+ATOM    258  N   VAL A  31      -6.835  -6.090   1.610  1.00 86.60           N
+ATOM    259  CA  VAL A  31      -6.583  -4.828   0.922  1.00 86.60           C
+ATOM    260  C   VAL A  31      -6.313  -5.090  -0.558  1.00 86.60           C
+ATOM    261  CB  VAL A  31      -5.397  -4.064   1.553  1.00 86.60           C
+ATOM    262  O   VAL A  31      -5.684  -6.090  -0.912  1.00 86.60           O
+ATOM    263  CG1 VAL A  31      -4.102  -4.859   1.404  1.00 86.60           C
+ATOM    264  CG2 VAL A  31      -5.255  -2.682   0.919  1.00 86.60           C
+ATOM    265  N   THR A  32      -6.886  -4.281  -1.455  1.00 86.60           N
+ATOM    266  CA  THR A  32      -6.648  -4.386  -2.890  1.00 86.60           C
+ATOM    267  C   THR A  32      -5.875  -3.173  -3.400  1.00 86.60           C
+ATOM    268  CB  THR A  32      -7.971  -4.518  -3.667  1.00 86.60           C
+ATOM    269  O   THR A  32      -6.311  -2.033  -3.224  1.00 86.60           O
+ATOM    270  CG2 THR A  32      -7.715  -4.710  -5.159  1.00 86.60           C
+ATOM    271  OG1 THR A  32      -8.700  -5.647  -3.170  1.00 86.60           O
+ATOM    272  N   PHE A  33      -4.627  -3.379  -3.960  1.00 85.17           N
+ATOM    273  CA  PHE A  33      -3.825  -2.315  -4.552  1.00 85.17           C
+ATOM    274  C   PHE A  33      -4.096  -2.201  -6.048  1.00 85.17           C
+ATOM    275  CB  PHE A  33      -2.333  -2.565  -4.307  1.00 85.17           C
+ATOM    276  O   PHE A  33      -4.048  -3.198  -6.771  1.00 85.17           O
+ATOM    277  CG  PHE A  33      -1.918  -2.397  -2.870  1.00 85.17           C
+ATOM    278  CD1 PHE A  33      -1.696  -1.132  -2.340  1.00 85.17           C
+ATOM    279  CD2 PHE A  33      -1.750  -3.504  -2.050  1.00 85.17           C
+ATOM    280  CE1 PHE A  33      -1.312  -0.973  -1.011  1.00 85.17           C
+ATOM    281  CE2 PHE A  33      -1.366  -3.354  -0.720  1.00 85.17           C
+ATOM    282  CZ  PHE A  33      -1.146  -2.087  -0.203  1.00 85.17           C
+ATOM    283  N   VAL A  34      -4.488  -1.033  -6.377  1.00 79.71           N
+ATOM    284  CA  VAL A  34      -4.738  -0.771  -7.791  1.00 79.71           C
+ATOM    285  C   VAL A  34      -3.582   0.033  -8.380  1.00 79.71           C
+ATOM    286  CB  VAL A  34      -6.072  -0.020  -8.002  1.00 79.71           C
+ATOM    287  O   VAL A  34      -3.285   1.137  -7.917  1.00 79.71           O
+ATOM    288  CG1 VAL A  34      -6.363   0.156  -9.491  1.00 79.71           C
+ATOM    289  CG2 VAL A  34      -7.216  -0.762  -7.314  1.00 79.71           C
+ATOM    290  N   ILE A  35      -2.823  -0.596  -9.337  1.00 77.12           N
+ATOM    291  CA  ILE A  35      -1.721   0.099  -9.993  1.00 77.12           C
+ATOM    292  C   ILE A  35      -2.004   0.218 -11.489  1.00 77.12           C
+ATOM    293  CB  ILE A  35      -0.376  -0.624  -9.757  1.00 77.12           C
+ATOM    294  O   ILE A  35      -2.768  -0.572 -12.047  1.00 77.12           O
+ATOM    295  CG1 ILE A  35      -0.441  -2.059 -10.291  1.00 77.12           C
+ATOM    296  CG2 ILE A  35      -0.007  -0.608  -8.271  1.00 77.12           C
+ATOM    297  CD1 ILE A  35       0.916  -2.740 -10.402  1.00 77.12           C
+ATOM    298  N   SER A  36      -1.599   1.325 -11.947  1.00 72.93           N
+ATOM    299  CA  SER A  36      -1.731   1.517 -13.387  1.00 72.93           C
+ATOM    300  C   SER A  36      -0.484   1.040 -14.125  1.00 72.93           C
+ATOM    301  CB  SER A  36      -1.994   2.988 -13.711  1.00 72.93           C
+ATOM    302  O   SER A  36       0.613   1.040 -13.564  1.00 72.93           O
+ATOM    303  OG  SER A  36      -1.967   3.208 -15.110  1.00 72.93           O
+ATOM    304  N   SER A  37      -0.670   0.403 -15.236  1.00 69.39           N
+ATOM    305  CA  SER A  37       0.426  -0.073 -16.073  1.00 69.39           C
+ATOM    306  C   SER A  37       1.372   1.064 -16.443  1.00 69.39           C
+ATOM    307  CB  SER A  37      -0.115  -0.731 -17.343  1.00 69.39           C
+ATOM    308  O   SER A  37       2.505   0.823 -16.867  1.00 69.39           O
+ATOM    309  OG  SER A  37      -1.045   0.119 -17.992  1.00 69.39           O
+ATOM    310  N   ALA A  38       0.862   2.297 -16.248  1.00 67.39           N
+ATOM    311  CA  ALA A  38       1.656   3.492 -16.525  1.00 67.39           C
+ATOM    312  C   ALA A  38       2.561   3.834 -15.344  1.00 67.39           C
+ATOM    313  CB  ALA A  38       0.745   4.672 -16.853  1.00 67.39           C
+ATOM    314  O   ALA A  38       3.386   4.746 -15.431  1.00 67.39           O
+ATOM    315  N   ASP A  39       2.279   3.071 -14.251  1.00 72.75           N
+ATOM    316  CA  ASP A  39       3.057   3.298 -13.037  1.00 72.75           C
+ATOM    317  C   ASP A  39       4.501   2.833 -13.215  1.00 72.75           C
+ATOM    318  CB  ASP A  39       2.417   2.580 -11.847  1.00 72.75           C
+ATOM    319  O   ASP A  39       4.780   1.959 -14.039  1.00 72.75           O
+ATOM    320  CG  ASP A  39       1.183   3.292 -11.320  1.00 72.75           C
+ATOM    321  OD1 ASP A  39       1.293   4.454 -10.874  1.00 72.75           O
+ATOM    322  OD2 ASP A  39       0.091   2.684 -11.348  1.00 72.75           O
+ATOM    323  N   SER A  40       5.475   3.562 -12.768  1.00 72.73           N
+ATOM    324  CA  SER A  40       6.908   3.286 -12.803  1.00 72.73           C
+ATOM    325  C   SER A  40       7.256   2.065 -11.958  1.00 72.73           C
+ATOM    326  CB  SER A  40       7.700   4.499 -12.314  1.00 72.73           C
+ATOM    327  O   SER A  40       6.491   1.677 -11.073  1.00 72.73           O
+ATOM    328  OG  SER A  40       7.410   5.640 -13.101  1.00 72.73           O
+ATOM    329  N   GLU A  41       8.265   1.200 -12.472  1.00 83.34           N
+ATOM    330  CA  GLU A  41       8.802   0.051 -11.749  1.00 83.34           C
+ATOM    331  C   GLU A  41       8.950   0.356 -10.261  1.00 83.34           C
+ATOM    332  CB  GLU A  41      10.152  -0.371 -12.336  1.00 83.34           C
+ATOM    333  O   GLU A  41       8.722  -0.515  -9.418  1.00 83.34           O
+ATOM    334  CG  GLU A  41      10.046  -1.058 -13.689  1.00 83.34           C
+ATOM    335  CD  GLU A  41      11.377  -1.586 -14.200  1.00 83.34           C
+ATOM    336  OE1 GLU A  41      11.690  -2.775 -13.962  1.00 83.34           O
+ATOM    337  OE2 GLU A  41      12.113  -0.805 -14.842  1.00 83.34           O
+ATOM    338  N   GLU A  42       9.164   1.541  -9.996  1.00 85.29           N
+ATOM    339  CA  GLU A  42       9.324   1.965  -8.608  1.00 85.29           C
+ATOM    340  C   GLU A  42       8.026   1.793  -7.825  1.00 85.29           C
+ATOM    341  CB  GLU A  42       9.789   3.422  -8.541  1.00 85.29           C
+ATOM    342  O   GLU A  42       8.031   1.257  -6.715  1.00 85.29           O
+ATOM    343  CG  GLU A  42      10.241   3.861  -7.156  1.00 85.29           C
+ATOM    344  CD  GLU A  42      11.177   5.059  -7.182  1.00 85.29           C
+ATOM    345  OE1 GLU A  42      10.689   6.208  -7.276  1.00 85.29           O
+ATOM    346  OE2 GLU A  42      12.408   4.847  -7.108  1.00 85.29           O
+ATOM    347  N   ILE A  43       6.916   2.343  -8.302  1.00 82.81           N
+ATOM    348  CA  ILE A  43       5.610   2.254  -7.658  1.00 82.81           C
+ATOM    349  C   ILE A  43       5.202   0.789  -7.519  1.00 82.81           C
+ATOM    350  CB  ILE A  43       4.537   3.036  -8.447  1.00 82.81           C
+ATOM    351  O   ILE A  43       4.717   0.370  -6.465  1.00 82.81           O
+ATOM    352  CG1 ILE A  43       4.822   4.541  -8.387  1.00 82.81           C
+ATOM    353  CG2 ILE A  43       3.137   2.722  -7.911  1.00 82.81           C
+ATOM    354  CD1 ILE A  43       3.876   5.384  -9.232  1.00 82.81           C
+ATOM    355  N   LYS A  44       5.549  -0.013  -8.535  1.00 83.23           N
+ATOM    356  CA  LYS A  44       5.294  -1.450  -8.502  1.00 83.23           C
+ATOM    357  C   LYS A  44       6.064  -2.118  -7.366  1.00 83.23           C
+ATOM    358  CB  LYS A  44       5.667  -2.093  -9.838  1.00 83.23           C
+ATOM    359  O   LYS A  44       5.511  -2.944  -6.636  1.00 83.23           O
+ATOM    360  CG  LYS A  44       4.624  -1.902 -10.929  1.00 83.23           C
+ATOM    361  CD  LYS A  44       4.985  -2.679 -12.189  1.00 83.23           C
+ATOM    362  CE  LYS A  44       4.004  -2.398 -13.320  1.00 83.23           C
+ATOM    363  NZ  LYS A  44       4.334  -3.183 -14.547  1.00 83.23           N
+ATOM    364  N   LYS A  45       7.296  -1.791  -7.304  1.00 88.48           N
+ATOM    365  CA  LYS A  45       8.146  -2.345  -6.255  1.00 88.48           C
+ATOM    366  C   LYS A  45       7.628  -1.967  -4.870  1.00 88.48           C
+ATOM    367  CB  LYS A  45       9.589  -1.867  -6.424  1.00 88.48           C
+ATOM    368  O   LYS A  45       7.601  -2.800  -3.962  1.00 88.48           O
+ATOM    369  CG  LYS A  45      10.363  -2.605  -7.506  1.00 88.48           C
+ATOM    370  CD  LYS A  45      11.819  -2.159  -7.554  1.00 88.48           C
+ATOM    371  CE  LYS A  45      12.582  -2.860  -8.670  1.00 88.48           C
+ATOM    372  NZ  LYS A  45      14.014  -2.437  -8.712  1.00 88.48           N
+ATOM    373  N   LEU A  46       7.267  -0.760  -4.707  1.00 90.20           N
+ATOM    374  CA  LEU A  46       6.764  -0.277  -3.426  1.00 90.20           C
+ATOM    375  C   LEU A  46       5.478  -1.001  -3.040  1.00 90.20           C
+ATOM    376  CB  LEU A  46       6.516   1.233  -3.481  1.00 90.20           C
+ATOM    377  O   LEU A  46       5.308  -1.401  -1.886  1.00 90.20           O
+ATOM    378  CG  LEU A  46       7.756   2.126  -3.413  1.00 90.20           C
+ATOM    379  CD1 LEU A  46       7.391   3.567  -3.755  1.00 90.20           C
+ATOM    380  CD2 LEU A  46       8.398   2.045  -2.033  1.00 90.20           C
+ATOM    381  N   VAL A  47       4.634  -1.055  -3.956  1.00 87.72           N
+ATOM    382  CA  VAL A  47       3.355  -1.724  -3.743  1.00 87.72           C
+ATOM    383  C   VAL A  47       3.592  -3.184  -3.363  1.00 87.72           C
+ATOM    384  CB  VAL A  47       2.454  -1.638  -4.995  1.00 87.72           C
+ATOM    385  O   VAL A  47       2.953  -3.705  -2.446  1.00 87.72           O
+ATOM    386  CG1 VAL A  47       1.201  -2.494  -4.819  1.00 87.72           C
+ATOM    387  CG2 VAL A  47       2.077  -0.186  -5.282  1.00 87.72           C
+ATOM    388  N   GLU A  48       4.453  -3.832  -4.059  1.00 89.12           N
+ATOM    389  CA  GLU A  48       4.817  -5.214  -3.761  1.00 89.12           C
+ATOM    390  C   GLU A  48       5.366  -5.347  -2.344  1.00 89.12           C
+ATOM    391  CB  GLU A  48       5.843  -5.729  -4.774  1.00 89.12           C
+ATOM    392  O   GLU A  48       5.015  -6.283  -1.622  1.00 89.12           O
+ATOM    393  CG  GLU A  48       5.260  -6.014  -6.150  1.00 89.12           C
+ATOM    394  CD  GLU A  48       6.211  -6.777  -7.058  1.00 89.12           C
+ATOM    395  OE1 GLU A  48       6.128  -8.025  -7.112  1.00 89.12           O
+ATOM    396  OE2 GLU A  48       7.046  -6.122  -7.721  1.00 89.12           O
+ATOM    397  N   GLU A  49       6.263  -4.525  -2.078  1.00 92.03           N
+ATOM    398  CA  GLU A  49       6.825  -4.532  -0.731  1.00 92.03           C
+ATOM    399  C   GLU A  49       5.738  -4.339   0.322  1.00 92.03           C
+ATOM    400  CB  GLU A  49       7.894  -3.445  -0.589  1.00 92.03           C
+ATOM    401  O   GLU A  49       5.730  -5.026   1.346  1.00 92.03           O
+ATOM    402  CG  GLU A  49       9.203  -3.773  -1.292  1.00 92.03           C
+ATOM    403  CD  GLU A  49      10.340  -2.840  -0.909  1.00 92.03           C
+ATOM    404  OE1 GLU A  49      10.970  -3.054   0.152  1.00 92.03           O
+ATOM    405  OE2 GLU A  49      10.603  -1.886  -1.675  1.00 92.03           O
+ATOM    406  N   ASN A  50       4.950  -3.304   0.106  1.00 90.05           N
+ATOM    407  CA  ASN A  50       3.830  -3.068   1.010  1.00 90.05           C
+ATOM    408  C   ASN A  50       2.960  -4.313   1.161  1.00 90.05           C
+ATOM    409  CB  ASN A  50       2.987  -1.887   0.524  1.00 90.05           C
+ATOM    410  O   ASN A  50       2.539  -4.651   2.269  1.00 90.05           O
+ATOM    411  CG  ASN A  50       3.539  -0.550   0.978  1.00 90.05           C
+ATOM    412  ND2 ASN A  50       4.277   0.116   0.098  1.00 90.05           N
+ATOM    413  OD1 ASN A  50       3.305  -0.120   2.110  1.00 90.05           O
+ATOM    414  N   ALA A  51       2.704  -4.904   0.020  1.00 89.43           N
+ATOM    415  CA  ALA A  51       1.897  -6.121  -0.004  1.00 89.43           C
+ATOM    416  C   ALA A  51       2.538  -7.220   0.839  1.00 89.43           C
+ATOM    417  CB  ALA A  51       1.701  -6.602  -1.439  1.00 89.43           C
+ATOM    418  O   ALA A  51       1.849  -7.920   1.585  1.00 89.43           O
+ATOM    419  N   GLU A  52       3.815  -7.414   0.696  1.00 91.12           N
+ATOM    420  CA  GLU A  52       4.547  -8.437   1.437  1.00 91.12           C
+ATOM    421  C   GLU A  52       4.436  -8.213   2.942  1.00 91.12           C
+ATOM    422  CB  GLU A  52       6.018  -8.457   1.015  1.00 91.12           C
+ATOM    423  O   GLU A  52       4.209  -9.158   3.700  1.00 91.12           O
+ATOM    424  CG  GLU A  52       6.248  -9.005  -0.387  1.00 91.12           C
+ATOM    425  CD  GLU A  52       7.720  -9.166  -0.733  1.00 91.12           C
+ATOM    426  OE1 GLU A  52       8.313 -10.215  -0.392  1.00 91.12           O
+ATOM    427  OE2 GLU A  52       8.285  -8.235  -1.350  1.00 91.12           O
+ATOM    428  N   ILE A  53       4.609  -7.008   3.402  1.00 91.89           N
+ATOM    429  CA  ILE A  53       4.531  -6.669   4.819  1.00 91.89           C
+ATOM    430  C   ILE A  53       3.112  -6.910   5.330  1.00 91.89           C
+ATOM    431  CB  ILE A  53       4.950  -5.204   5.074  1.00 91.89           C
+ATOM    432  O   ILE A  53       2.921  -7.486   6.403  1.00 91.89           O
+ATOM    433  CG1 ILE A  53       6.419  -4.994   4.686  1.00 91.89           C
+ATOM    434  CG2 ILE A  53       4.710  -4.819   6.536  1.00 91.89           C
+ATOM    435  CD1 ILE A  53       6.878  -3.544   4.762  1.00 91.89           C
+ATOM    436  N   LEU A  54       2.220  -6.422   4.598  1.00 90.60           N
+ATOM    437  CA  LEU A  54       0.817  -6.592   4.960  1.00 90.60           C
+ATOM    438  C   LEU A  54       0.454  -8.071   5.048  1.00 90.60           C
+ATOM    439  CB  LEU A  54      -0.087  -5.892   3.942  1.00 90.60           C
+ATOM    440  O   LEU A  54      -0.274  -8.482   5.955  1.00 90.60           O
+ATOM    441  CG  LEU A  54      -0.161  -4.367   4.036  1.00 90.60           C
+ATOM    442  CD1 LEU A  54      -0.837  -3.792   2.797  1.00 90.60           C
+ATOM    443  CD2 LEU A  54      -0.900  -3.944   5.301  1.00 90.60           C
+ATOM    444  N   GLU A  55       0.909  -8.858   4.086  1.00 90.94           N
+ATOM    445  CA  GLU A  55       0.715 -10.305   4.103  1.00 90.94           C
+ATOM    446  C   GLU A  55       1.303 -10.926   5.367  1.00 90.94           C
+ATOM    447  CB  GLU A  55       1.342 -10.946   2.862  1.00 90.94           C
+ATOM    448  O   GLU A  55       0.712 -11.837   5.950  1.00 90.94           O
+ATOM    449  CG  GLU A  55       0.549 -10.716   1.584  1.00 90.94           C
+ATOM    450  CD  GLU A  55       0.996 -11.603   0.432  1.00 90.94           C
+ATOM    451  OE1 GLU A  55       0.436 -12.710   0.266  1.00 90.94           O
+ATOM    452  OE2 GLU A  55       1.915 -11.187  -0.308  1.00 90.94           O
+ATOM    453  N   LYS A  56       2.477 -10.439   5.689  1.00 90.58           N
+ATOM    454  CA  LYS A  56       3.128 -10.914   6.907  1.00 90.58           C
+ATOM    455  C   LYS A  56       2.303 -10.564   8.143  1.00 90.58           C
+ATOM    456  CB  LYS A  56       4.533 -10.324   7.031  1.00 90.58           C
+ATOM    457  O   LYS A  56       2.287 -11.317   9.119  1.00 90.58           O
+ATOM    458  CG  LYS A  56       5.551 -10.945   6.085  1.00 90.58           C
+ATOM    459  CD  LYS A  56       6.949 -10.393   6.327  1.00 90.58           C
+ATOM    460  CE  LYS A  56       7.967 -11.008   5.376  1.00 90.58           C
+ATOM    461  NZ  LYS A  56       9.350 -10.516   5.651  1.00 90.58           N
+ATOM    462  N   LEU A  57       1.666  -9.472   8.081  1.00 89.28           N
+ATOM    463  CA  LEU A  57       0.843  -8.979   9.180  1.00 89.28           C
+ATOM    464  C   LEU A  57      -0.522  -9.661   9.185  1.00 89.28           C
+ATOM    465  CB  LEU A  57       0.668  -7.462   9.080  1.00 89.28           C
+ATOM    466  O   LEU A  57      -1.332  -9.430  10.085  1.00 89.28           O
+ATOM    467  CG  LEU A  57       1.906  -6.615   9.381  1.00 89.28           C
+ATOM    468  CD1 LEU A  57       1.651  -5.157   9.016  1.00 89.28           C
+ATOM    469  CD2 LEU A  57       2.298  -6.744  10.849  1.00 89.28           C
+ATOM    470  N   GLY A  58      -0.801 -10.562   8.254  1.00 89.24           N
+ATOM    471  CA  GLY A  58      -2.043 -11.316   8.202  1.00 89.24           C
+ATOM    472  C   GLY A  58      -3.109 -10.651   7.352  1.00 89.24           C
+ATOM    473  O   GLY A  58      -4.304 -10.868   7.563  1.00 89.24           O
+ATOM    474  N   VAL A  59      -2.742  -9.745   6.559  1.00 90.59           N
+ATOM    475  CA  VAL A  59      -3.658  -9.015   5.689  1.00 90.59           C
+ATOM    476  C   VAL A  59      -3.626  -9.615   4.285  1.00 90.59           C
+ATOM    477  CB  VAL A  59      -3.312  -7.511   5.634  1.00 90.59           C
+ATOM    478  O   VAL A  59      -2.551  -9.854   3.730  1.00 90.59           O
+ATOM    479  CG1 VAL A  59      -4.305  -6.760   4.750  1.00 90.59           C
+ATOM    480  CG2 VAL A  59      -3.290  -6.917   7.042  1.00 90.59           C
+ATOM    481  N   LYS A  60      -4.796 -10.038   3.669  1.00 91.31           N
+ATOM    482  CA  LYS A  60      -4.890 -10.525   2.295  1.00 91.31           C
+ATOM    483  C   LYS A  60      -4.751  -9.380   1.296  1.00 91.31           C
+ATOM    484  CB  LYS A  60      -6.215 -11.256   2.074  1.00 91.31           C
+ATOM    485  O   LYS A  60      -5.554  -8.444   1.300  1.00 91.31           O
+ATOM    486  CG  LYS A  60      -6.280 -12.039   0.771  1.00 91.31           C
+ATOM    487  CD  LYS A  60      -7.562 -12.856   0.673  1.00 91.31           C
+ATOM    488  CE  LYS A  60      -7.618 -13.657  -0.621  1.00 91.31           C
+ATOM    489  NZ  LYS A  60      -8.859 -14.483  -0.707  1.00 91.31           N
+ATOM    490  N   VAL A  61      -3.640  -9.388   0.542  1.00 90.50           N
+ATOM    491  CA  VAL A  61      -3.325  -8.295  -0.373  1.00 90.50           C
+ATOM    492  C   VAL A  61      -3.627  -8.719  -1.808  1.00 90.50           C
+ATOM    493  CB  VAL A  61      -1.849  -7.856  -0.246  1.00 90.50           C
+ATOM    494  O   VAL A  61      -3.177  -9.776  -2.258  1.00 90.50           O
+ATOM    495  CG1 VAL A  61      -1.548  -6.696  -1.192  1.00 90.50           C
+ATOM    496  CG2 VAL A  61      -1.531  -7.470   1.197  1.00 90.50           C
+ATOM    497  N   LYS A  62      -4.464  -7.942  -2.568  1.00 89.89           N
+ATOM    498  CA  LYS A  62      -4.755  -8.135  -3.986  1.00 89.89           C
+ATOM    499  C   LYS A  62      -4.271  -6.948  -4.813  1.00 89.89           C
+ATOM    500  CB  LYS A  62      -6.254  -8.348  -4.202  1.00 89.89           C
+ATOM    501  O   LYS A  62      -4.598  -5.799  -4.510  1.00 89.89           O
+ATOM    502  CG  LYS A  62      -6.615  -8.835  -5.598  1.00 89.89           C
+ATOM    503  CD  LYS A  62      -8.100  -9.156  -5.711  1.00 89.89           C
+ATOM    504  CE  LYS A  62      -8.454  -9.686  -7.094  1.00 89.89           C
+ATOM    505  NZ  LYS A  62      -9.906 -10.018  -7.205  1.00 89.89           N
+ATOM    506  N   VAL A  63      -3.369  -7.174  -5.795  1.00 85.66           N
+ATOM    507  CA  VAL A  63      -2.847  -6.112  -6.649  1.00 85.66           C
+ATOM    508  C   VAL A  63      -3.490  -6.198  -8.032  1.00 85.66           C
+ATOM    509  CB  VAL A  63      -1.309  -6.185  -6.772  1.00 85.66           C
+ATOM    510  O   VAL A  63      -3.359  -7.211  -8.723  1.00 85.66           O
+ATOM    511  CG1 VAL A  63      -0.779  -5.034  -7.625  1.00 85.66           C
+ATOM    512  CG2 VAL A  63      -0.663  -6.171  -5.388  1.00 85.66           C
+ATOM    513  N   VAL A  64      -4.304  -5.230  -8.309  1.00 83.17           N
+ATOM    514  CA  VAL A  64      -5.001  -5.174  -9.589  1.00 83.17           C
+ATOM    515  C   VAL A  64      -4.339  -4.137 -10.494  1.00 83.17           C
+ATOM    516  CB  VAL A  64      -6.499  -4.843  -9.406  1.00 83.17           C
+ATOM    517  O   VAL A  64      -4.094  -3.004 -10.073  1.00 83.17           O
+ATOM    518  CG1 VAL A  64      -7.221  -4.843 -10.753  1.00 83.17           C
+ATOM    519  CG2 VAL A  64      -7.152  -5.836  -8.447  1.00 83.17           C
+ATOM    520  N   GLU A  65      -3.914  -4.576 -11.655  1.00 82.38           N
+ATOM    521  CA  GLU A  65      -3.300  -3.697 -12.646  1.00 82.38           C
+ATOM    522  C   GLU A  65      -4.341  -3.154 -13.621  1.00 82.38           C
+ATOM    523  CB  GLU A  65      -2.198  -4.434 -13.410  1.00 82.38           C
+ATOM    524  O   GLU A  65      -5.007  -3.922 -14.318  1.00 82.38           O
+ATOM    525  CG  GLU A  65      -1.357  -3.531 -14.300  1.00 82.38           C
+ATOM    526  CD  GLU A  65      -0.228  -4.266 -15.005  1.00 82.38           C
+ATOM    527  OE1 GLU A  65      -0.445  -4.784 -16.124  1.00 82.38           O
+ATOM    528  OE2 GLU A  65       0.884  -4.325 -14.433  1.00 82.38           O
+ATOM    529  N   LEU A  66      -4.731  -1.848 -13.511  1.00 73.61           N
+ATOM    530  CA  LEU A  66      -5.704  -1.232 -14.407  1.00 73.61           C
+ATOM    531  C   LEU A  66      -5.031  -0.737 -15.682  1.00 73.61           C
+ATOM    532  CB  LEU A  66      -6.417  -0.072 -13.708  1.00 73.61           C
+ATOM    533  O   LEU A  66      -3.927  -0.189 -15.634  1.00 73.61           O
+ATOM    534  CG  LEU A  66      -7.425  -0.450 -12.622  1.00 73.61           C
+ATOM    535  CD1 LEU A  66      -7.847   0.787 -11.836  1.00 73.61           C
+ATOM    536  CD2 LEU A  66      -8.638  -1.141 -13.234  1.00 73.61           C
+ATOM    537  N   GLY A  67      -5.215  -1.540 -16.831  1.00 61.81           N
+ATOM    538  CA  GLY A  67      -4.761  -1.145 -18.155  1.00 61.81           C
+ATOM    539  C   GLY A  67      -5.505   0.056 -18.708  1.00 61.81           C
+ATOM    540  O   GLY A  67      -6.738   0.065 -18.746  1.00 61.81           O
+ATOM    541  N   GLY A  68      -5.090   1.390 -18.491  1.00 60.96           N
+ATOM    542  CA  GLY A  68      -5.697   2.625 -18.962  1.00 60.96           C
+ATOM    543  C   GLY A  68      -4.768   3.820 -18.865  1.00 60.96           C
+ATOM    544  O   GLY A  68      -3.738   3.760 -18.191  1.00 60.96           O
+ATOM    545  N   GLU A  69      -4.531   4.607 -19.986  1.00 63.19           N
+ATOM    546  CA  GLU A  69      -3.769   5.840 -20.158  1.00 63.19           C
+ATOM    547  C   GLU A  69      -3.863   6.725 -18.918  1.00 63.19           C
+ATOM    548  CB  GLU A  69      -4.257   6.607 -21.389  1.00 63.19           C
+ATOM    549  O   GLU A  69      -4.938   6.870 -18.332  1.00 63.19           O
+ATOM    550  CG  GLU A  69      -3.142   7.015 -22.341  1.00 63.19           C
+ATOM    551  CD  GLU A  69      -3.653   7.632 -23.634  1.00 63.19           C
+ATOM    552  OE1 GLU A  69      -3.936   8.852 -23.654  1.00 63.19           O
+ATOM    553  OE2 GLU A  69      -3.771   6.890 -24.634  1.00 63.19           O
+ATOM    554  N   LEU A  70      -2.763   6.805 -18.127  1.00 62.85           N
+ATOM    555  CA  LEU A  70      -2.656   7.713 -16.990  1.00 62.85           C
+ATOM    556  C   LEU A  70      -3.013   9.139 -17.398  1.00 62.85           C
+ATOM    557  CB  LEU A  70      -1.242   7.676 -16.406  1.00 62.85           C
+ATOM    558  O   LEU A  70      -2.495   9.653 -18.391  1.00 62.85           O
+ATOM    559  CG  LEU A  70      -0.896   6.460 -15.544  1.00 62.85           C
+ATOM    560  CD1 LEU A  70       0.611   6.230 -15.534  1.00 62.85           C
+ATOM    561  CD2 LEU A  70      -1.425   6.642 -14.126  1.00 62.85           C
+ATOM    562  N   ASP A  71      -4.382   9.504 -17.115  1.00 67.78           N
+ATOM    563  CA  ASP A  71      -4.779  10.889 -17.347  1.00 67.78           C
+ATOM    564  C   ASP A  71      -3.796  11.859 -16.695  1.00 67.78           C
+ATOM    565  CB  ASP A  71      -6.193  11.139 -16.819  1.00 67.78           C
+ATOM    566  O   ASP A  71      -3.070  11.487 -15.770  1.00 67.78           O
+ATOM    567  CG  ASP A  71      -6.996  12.083 -17.696  1.00 67.78           C
+ATOM    568  OD1 ASP A  71      -7.054  13.295 -17.394  1.00 67.78           O
+ATOM    569  OD2 ASP A  71      -7.578  11.611 -18.697  1.00 67.78           O
+ATOM    570  N   GLU A  72      -3.069  12.841 -17.280  1.00 66.31           N
+ATOM    571  CA  GLU A  72      -2.192  13.936 -16.876  1.00 66.31           C
+ATOM    572  C   GLU A  72      -2.299  14.205 -15.378  1.00 66.31           C
+ATOM    573  CB  GLU A  72      -2.521  15.207 -17.663  1.00 66.31           C
+ATOM    574  O   GLU A  72      -1.296  14.487 -14.719  1.00 66.31           O
+ATOM    575  CG  GLU A  72      -1.767  15.327 -18.980  1.00 66.31           C
+ATOM    576  CD  GLU A  72      -2.109  16.590 -19.754  1.00 66.31           C
+ATOM    577  OE1 GLU A  72      -1.474  17.642 -19.511  1.00 66.31           O
+ATOM    578  OE2 GLU A  72      -3.020  16.529 -20.609  1.00 66.31           O
+ATOM    579  N   LEU A  73      -3.365  13.942 -14.773  1.00 66.33           N
+ATOM    580  CA  LEU A  73      -3.564  14.142 -13.342  1.00 66.33           C
+ATOM    581  C   LEU A  73      -2.893  13.032 -12.539  1.00 66.33           C
+ATOM    582  CB  LEU A  73      -5.058  14.196 -13.010  1.00 66.33           C
+ATOM    583  O   LEU A  73      -2.291  13.292 -11.495  1.00 66.33           O
+ATOM    584  CG  LEU A  73      -5.727  15.567 -13.115  1.00 66.33           C
+ATOM    585  CD1 LEU A  73      -7.205  15.411 -13.456  1.00 66.33           C
+ATOM    586  CD2 LEU A  73      -5.551  16.349 -11.818  1.00 66.33           C
+ATOM    587  N   THR A  74      -2.894  11.771 -13.049  1.00 65.78           N
+ATOM    588  CA  THR A  74      -2.225  10.632 -12.430  1.00 65.78           C
+ATOM    589  C   THR A  74      -0.710  10.750 -12.575  1.00 65.78           C
+ATOM    590  CB  THR A  74      -2.699   9.303 -13.046  1.00 65.78           C
+ATOM    591  O   THR A  74       0.035  10.429 -11.646  1.00 65.78           O
+ATOM    592  CG2 THR A  74      -2.057   8.111 -12.344  1.00 65.78           C
+ATOM    593  OG1 THR A  74      -4.124   9.206 -12.920  1.00 65.78           O
+ATOM    594  N   LEU A  75      -0.311  11.344 -13.746  1.00 68.90           N
+ATOM    595  CA  LEU A  75       1.107  11.593 -13.982  1.00 68.90           C
+ATOM    596  C   LEU A  75       1.625  12.698 -13.068  1.00 68.90           C
+ATOM    597  CB  LEU A  75       1.348  11.970 -15.446  1.00 68.90           C
+ATOM    598  O   LEU A  75       2.740  12.610 -12.550  1.00 68.90           O
+ATOM    599  CG  LEU A  75       1.401  10.812 -16.445  1.00 68.90           C
+ATOM    600  CD1 LEU A  75       1.017  11.297 -17.839  1.00 68.90           C
+ATOM    601  CD2 LEU A  75       2.788  10.179 -16.456  1.00 68.90           C
+ATOM    602  N   LYS A  76       0.708  13.698 -12.805  1.00 71.32           N
+ATOM    603  CA  LYS A  76       1.071  14.813 -11.936  1.00 71.32           C
+ATOM    604  C   LYS A  76       1.146  14.372 -10.477  1.00 71.32           C
+ATOM    605  CB  LYS A  76       0.068  15.959 -12.086  1.00 71.32           C
+ATOM    606  O   LYS A  76       1.991  14.854  -9.720  1.00 71.32           O
+ATOM    607  CG  LYS A  76       0.514  17.260 -11.436  1.00 71.32           C
+ATOM    608  CD  LYS A  76      -0.488  18.381 -11.684  1.00 71.32           C
+ATOM    609  CE  LYS A  76      -0.066  19.672 -10.996  1.00 71.32           C
+ATOM    610  NZ  LYS A  76      -1.069  20.760 -11.197  1.00 71.32           N
+ATOM    611  N   HIS A  77       0.420  13.365 -10.189  1.00 75.38           N
+ATOM    612  CA  HIS A  77       0.315  12.913  -8.807  1.00 75.38           C
+ATOM    613  C   HIS A  77       1.167  11.672  -8.567  1.00 75.38           C
+ATOM    614  CB  HIS A  77      -1.144  12.626  -8.446  1.00 75.38           C
+ATOM    615  O   HIS A  77       1.146  11.101  -7.474  1.00 75.38           O
+ATOM    616  CG  HIS A  77      -1.983  13.857  -8.316  1.00 75.38           C
+ATOM    617  CD2 HIS A  77      -1.961  14.855  -7.401  1.00 75.38           C
+ATOM    618  ND1 HIS A  77      -2.994  14.166  -9.199  1.00 75.38           N
+ATOM    619  CE1 HIS A  77      -3.559  15.304  -8.833  1.00 75.38           C
+ATOM    620  NE2 HIS A  77      -2.951  15.743  -7.744  1.00 75.38           N
+ATOM    621  N   LYS A  78       1.936  11.316  -9.653  1.00 73.17           N
+ATOM    622  CA  LYS A  78       2.824  10.158  -9.610  1.00 73.17           C
+ATOM    623  C   LYS A  78       3.862  10.302  -8.500  1.00 73.17           C
+ATOM    624  CB  LYS A  78       3.521   9.965 -10.957  1.00 73.17           C
+ATOM    625  O   LYS A  78       4.117   9.353  -7.755  1.00 73.17           O
+ATOM    626  CG  LYS A  78       4.245   8.633 -11.095  1.00 73.17           C
+ATOM    627  CD  LYS A  78       4.838   8.461 -12.487  1.00 73.17           C
+ATOM    628  CE  LYS A  78       5.641   7.172 -12.598  1.00 73.17           C
+ATOM    629  NZ  LYS A  78       6.222   6.996 -13.963  1.00 73.17           N
+ATOM    630  N   GLU A  79       4.467  11.425  -8.459  1.00 82.01           N
+ATOM    631  CA  GLU A  79       5.451  11.676  -7.410  1.00 82.01           C
+ATOM    632  C   GLU A  79       4.810  11.619  -6.027  1.00 82.01           C
+ATOM    633  CB  GLU A  79       6.127  13.034  -7.621  1.00 82.01           C
+ATOM    634  O   GLU A  79       5.388  11.062  -5.091  1.00 82.01           O
+ATOM    635  CG  GLU A  79       7.125  13.052  -8.769  1.00 82.01           C
+ATOM    636  CD  GLU A  79       7.967  14.317  -8.812  1.00 82.01           C
+ATOM    637  OE1 GLU A  79       9.088  14.320  -8.254  1.00 82.01           O
+ATOM    638  OE2 GLU A  79       7.502  15.314  -9.408  1.00 82.01           O
+ATOM    639  N   LYS A  80       3.748  12.178  -5.930  1.00 81.73           N
+ATOM    640  CA  LYS A  80       3.031  12.190  -4.659  1.00 81.73           C
+ATOM    641  C   LYS A  80       2.604  10.781  -4.255  1.00 81.73           C
+ATOM    642  CB  LYS A  80       1.808  13.105  -4.740  1.00 81.73           C
+ATOM    643  O   LYS A  80       2.693  10.412  -3.082  1.00 81.73           O
+ATOM    644  CG  LYS A  80       2.129  14.583  -4.570  1.00 81.73           C
+ATOM    645  CD  LYS A  80       0.863  15.427  -4.516  1.00 81.73           C
+ATOM    646  CE  LYS A  80       1.184  16.914  -4.444  1.00 81.73           C
+ATOM    647  NZ  LYS A  80      -0.055  17.744  -4.351  1.00 81.73           N
+ATOM    648  N   ILE A  81       2.180  10.044  -5.166  1.00 80.10           N
+ATOM    649  CA  ILE A  81       1.790   8.655  -4.957  1.00 80.10           C
+ATOM    650  C   ILE A  81       2.993   7.848  -4.473  1.00 80.10           C
+ATOM    651  CB  ILE A  81       1.208   8.030  -6.245  1.00 80.10           C
+ATOM    652  O   ILE A  81       2.875   7.039  -3.550  1.00 80.10           O
+ATOM    653  CG1 ILE A  81      -0.130   8.688  -6.600  1.00 80.10           C
+ATOM    654  CG2 ILE A  81       1.050   6.515  -6.087  1.00 80.10           C
+ATOM    655  CD1 ILE A  81      -0.679   8.279  -7.960  1.00 80.10           C
+ATOM    656  N   LEU A  82       4.081   8.122  -5.134  1.00 83.52           N
+ATOM    657  CA  LEU A  82       5.328   7.478  -4.739  1.00 83.52           C
+ATOM    658  C   LEU A  82       5.679   7.813  -3.294  1.00 83.52           C
+ATOM    659  CB  LEU A  82       6.470   7.905  -5.666  1.00 83.52           C
+ATOM    660  O   LEU A  82       5.984   6.919  -2.501  1.00 83.52           O
+ATOM    661  CG  LEU A  82       7.813   7.204  -5.456  1.00 83.52           C
+ATOM    662  CD1 LEU A  82       7.738   5.760  -5.939  1.00 83.52           C
+ATOM    663  CD2 LEU A  82       8.927   7.957  -6.174  1.00 83.52           C
+ATOM    664  N   GLU A  83       5.694   9.021  -2.968  1.00 87.96           N
+ATOM    665  CA  GLU A  83       6.011   9.477  -1.618  1.00 87.96           C
+ATOM    666  C   GLU A  83       5.056   8.873  -0.593  1.00 87.96           C
+ATOM    667  CB  GLU A  83       5.967  11.005  -1.542  1.00 87.96           C
+ATOM    668  O   GLU A  83       5.488   8.386   0.454  1.00 87.96           O
+ATOM    669  CG  GLU A  83       6.496  11.571  -0.233  1.00 87.96           C
+ATOM    670  CD  GLU A  83       6.453  13.090  -0.175  1.00 87.96           C
+ATOM    671  OE1 GLU A  83       5.424  13.652   0.264  1.00 87.96           O
+ATOM    672  OE2 GLU A  83       7.456  13.723  -0.573  1.00 87.96           O
+ATOM    673  N   LYS A  84       3.752   8.925  -0.931  1.00 87.17           N
+ATOM    674  CA  LYS A  84       2.749   8.416   0.000  1.00 87.17           C
+ATOM    675  C   LYS A  84       2.865   6.903   0.160  1.00 87.17           C
+ATOM    676  CB  LYS A  84       1.342   8.787  -0.470  1.00 87.17           C
+ATOM    677  O   LYS A  84       2.708   6.376   1.264  1.00 87.17           O
+ATOM    678  CG  LYS A  84       1.031  10.274  -0.383  1.00 87.17           C
+ATOM    679  CD  LYS A  84       0.932  10.739   1.064  1.00 87.17           C
+ATOM    680  CE  LYS A  84       0.563  12.214   1.154  1.00 87.17           C
+ATOM    681  NZ  LYS A  84       0.580  12.705   2.564  1.00 87.17           N
+ATOM    682  N   THR A  85       3.075   6.194  -0.878  1.00 86.48           N
+ATOM    683  CA  THR A  85       3.277   4.749  -0.859  1.00 86.48           C
+ATOM    684  C   THR A  85       4.478   4.381   0.007  1.00 86.48           C
+ATOM    685  CB  THR A  85       3.478   4.195  -2.282  1.00 86.48           C
+ATOM    686  O   THR A  85       4.438   3.396   0.747  1.00 86.48           O
+ATOM    687  CG2 THR A  85       3.571   2.673  -2.271  1.00 86.48           C
+ATOM    688  OG1 THR A  85       2.371   4.589  -3.103  1.00 86.48           O
+ATOM    689  N   LYS A  86       5.560   5.179  -0.120  1.00 89.61           N
+ATOM    690  CA  LYS A  86       6.738   4.988   0.722  1.00 89.61           C
+ATOM    691  C   LYS A  86       6.390   5.146   2.199  1.00 89.61           C
+ATOM    692  CB  LYS A  86       7.840   5.975   0.335  1.00 89.61           C
+ATOM    693  O   LYS A  86       6.864   4.379   3.040  1.00 89.61           O
+ATOM    694  CG  LYS A  86       8.596   5.594  -0.930  1.00 89.61           C
+ATOM    695  CD  LYS A  86       9.745   6.555  -1.205  1.00 89.61           C
+ATOM    696  CE  LYS A  86      10.475   6.201  -2.494  1.00 89.61           C
+ATOM    697  NZ  LYS A  86      11.592   7.151  -2.777  1.00 89.61           N
+ATOM    698  N   LEU A  87       5.725   6.133   2.442  1.00 90.63           N
+ATOM    699  CA  LEU A  87       5.303   6.398   3.814  1.00 90.63           C
+ATOM    700  C   LEU A  87       4.520   5.217   4.379  1.00 90.63           C
+ATOM    701  CB  LEU A  87       4.449   7.667   3.877  1.00 90.63           C
+ATOM    702  O   LEU A  87       4.736   4.815   5.524  1.00 90.63           O
+ATOM    703  CG  LEU A  87       4.127   8.198   5.274  1.00 90.63           C
+ATOM    704  CD1 LEU A  87       5.400   8.666   5.971  1.00 90.63           C
+ATOM    705  CD2 LEU A  87       3.108   9.330   5.195  1.00 90.63           C
+ATOM    706  N   VAL A  88       3.611   4.770   3.617  1.00 89.50           N
+ATOM    707  CA  VAL A  88       2.818   3.610   4.012  1.00 89.50           C
+ATOM    708  C   VAL A  88       3.741   2.436   4.330  1.00 89.50           C
+ATOM    709  CB  VAL A  88       1.808   3.212   2.913  1.00 89.50           C
+ATOM    710  O   VAL A  88       3.547   1.739   5.328  1.00 89.50           O
+ATOM    711  CG1 VAL A  88       1.118   1.895   3.265  1.00 89.50           C
+ATOM    712  CG2 VAL A  88       0.777   4.321   2.708  1.00 89.50           C
+ATOM    713  N   LEU A  89       4.633   2.181   3.505  1.00 90.95           N
+ATOM    714  CA  LEU A  89       5.629   1.136   3.718  1.00 90.95           C
+ATOM    715  C   LEU A  89       6.348   1.334   5.048  1.00 90.95           C
+ATOM    716  CB  LEU A  89       6.644   1.123   2.572  1.00 90.95           C
+ATOM    717  O   LEU A  89       6.528   0.380   5.809  1.00 90.95           O
+ATOM    718  CG  LEU A  89       7.716   0.033   2.629  1.00 90.95           C
+ATOM    719  CD1 LEU A  89       7.069  -1.348   2.620  1.00 90.95           C
+ATOM    720  CD2 LEU A  89       8.691   0.180   1.466  1.00 90.95           C
+ATOM    721  N   GLU A  90       6.858   2.505   5.268  1.00 91.77           N
+ATOM    722  CA  GLU A  90       7.540   2.813   6.521  1.00 91.77           C
+ATOM    723  C   GLU A  90       6.648   2.516   7.723  1.00 91.77           C
+ATOM    724  CB  GLU A  90       7.985   4.278   6.545  1.00 91.77           C
+ATOM    725  O   GLU A  90       7.113   1.982   8.731  1.00 91.77           O
+ATOM    726  CG  GLU A  90       9.127   4.590   5.589  1.00 91.77           C
+ATOM    727  CD  GLU A  90      10.319   5.241   6.271  1.00 91.77           C
+ATOM    728  OE1 GLU A  90      11.320   4.540   6.544  1.00 91.77           O
+ATOM    729  OE2 GLU A  90      10.252   6.462   6.535  1.00 91.77           O
+ATOM    730  N   ILE A  91       5.411   2.885   7.569  1.00 90.84           N
+ATOM    731  CA  ILE A  91       4.447   2.638   8.635  1.00 90.84           C
+ATOM    732  C   ILE A  91       4.287   1.135   8.849  1.00 90.84           C
+ATOM    733  CB  ILE A  91       3.078   3.283   8.321  1.00 90.84           C
+ATOM    734  O   ILE A  91       4.286   0.660   9.987  1.00 90.84           O
+ATOM    735  CG1 ILE A  91       3.208   4.810   8.263  1.00 90.84           C
+ATOM    736  CG2 ILE A  91       2.031   2.861   9.356  1.00 90.84           C
+ATOM    737  CD1 ILE A  91       1.989   5.514   7.684  1.00 90.84           C
+ATOM    738  N   LEU A  92       4.083   0.438   7.825  1.00 90.22           N
+ATOM    739  CA  LEU A  92       3.929  -1.012   7.890  1.00 90.22           C
+ATOM    740  C   LEU A  92       5.163  -1.662   8.506  1.00 90.22           C
+ATOM    741  CB  LEU A  92       3.677  -1.587   6.494  1.00 90.22           C
+ATOM    742  O   LEU A  92       5.045  -2.579   9.322  1.00 90.22           O
+ATOM    743  CG  LEU A  92       2.300  -1.314   5.887  1.00 90.22           C
+ATOM    744  CD1 LEU A  92       2.262  -1.772   4.432  1.00 90.22           C
+ATOM    745  CD2 LEU A  92       1.211  -2.006   6.699  1.00 90.22           C
+ATOM    746  N   LYS A  93       6.355  -1.270   8.049  1.00 89.96           N
+ATOM    747  CA  LYS A  93       7.611  -1.791   8.582  1.00 89.96           C
+ATOM    748  C   LYS A  93       7.703  -1.568  10.089  1.00 89.96           C
+ATOM    749  CB  LYS A  93       8.803  -1.137   7.881  1.00 89.96           C
+ATOM    750  O   LYS A  93       8.149  -2.450  10.826  1.00 89.96           O
+ATOM    751  CG  LYS A  93       9.068  -1.672   6.481  1.00 89.96           C
+ATOM    752  CD  LYS A  93      10.333  -1.070   5.884  1.00 89.96           C
+ATOM    753  CE  LYS A  93      10.606  -1.615   4.488  1.00 89.96           C
+ATOM    754  NZ  LYS A  93      11.842  -1.022   3.894  1.00 89.96           N
+ATOM    755  N   GLU A  94       7.401  -0.356  10.430  1.00 91.25           N
+ATOM    756  CA  GLU A  94       7.387  -0.032  11.853  1.00 91.25           C
+ATOM    757  C   GLU A  94       6.457  -0.965  12.623  1.00 91.25           C
+ATOM    758  CB  GLU A  94       6.968   1.425  12.069  1.00 91.25           C
+ATOM    759  O   GLU A  94       6.797  -1.429  13.713  1.00 91.25           O
+ATOM    760  CG  GLU A  94       8.067   2.432  11.765  1.00 91.25           C
+ATOM    761  CD  GLU A  94       7.783   3.817  12.324  1.00 91.25           C
+ATOM    762  OE1 GLU A  94       8.221   4.116  13.459  1.00 91.25           O
+ATOM    763  OE2 GLU A  94       7.116   4.610  11.623  1.00 91.25           O
+ATOM    764  N   ARG A  95       5.355  -1.225  12.075  1.00 85.31           N
+ATOM    765  CA  ARG A  95       4.398  -2.127  12.709  1.00 85.31           C
+ATOM    766  C   ARG A  95       4.938  -3.553  12.756  1.00 85.31           C
+ATOM    767  CB  ARG A  95       3.059  -2.098  11.969  1.00 85.31           C
+ATOM    768  O   ARG A  95       4.712  -4.276  13.728  1.00 85.31           O
+ATOM    769  CG  ARG A  95       2.304  -0.786  12.115  1.00 85.31           C
+ATOM    770  CD  ARG A  95       1.648  -0.662  13.483  1.00 85.31           C
+ATOM    771  NE  ARG A  95       0.780   0.509  13.560  1.00 85.31           N
+ATOM    772  NH1 ARG A  95      -1.032  -0.634  14.424  1.00 85.31           N
+ATOM    773  NH2 ARG A  95      -1.174   1.620  14.033  1.00 85.31           N
+ATOM    774  CZ  ARG A  95      -0.474   0.496  14.006  1.00 85.31           C
+ATOM    775  N   LEU A  96       5.431  -3.928  11.737  1.00 87.53           N
+ATOM    776  CA  LEU A  96       6.039  -5.251  11.646  1.00 87.53           C
+ATOM    777  C   LEU A  96       7.142  -5.415  12.687  1.00 87.53           C
+ATOM    778  CB  LEU A  96       6.605  -5.486  10.244  1.00 87.53           C
+ATOM    779  O   LEU A  96       7.260  -6.471  13.311  1.00 87.53           O
+ATOM    780  CG  LEU A  96       7.106  -6.899   9.943  1.00 87.53           C
+ATOM    781  CD1 LEU A  96       5.938  -7.879   9.903  1.00 87.53           C
+ATOM    782  CD2 LEU A  96       7.877  -6.925   8.627  1.00 87.53           C
+ATOM    783  N   LYS A  97       8.035  -4.458  12.855  1.00 87.68           N
+ATOM    784  CA  LYS A  97       9.111  -4.470  13.842  1.00 87.68           C
+ATOM    785  C   LYS A  97       8.555  -4.573  15.260  1.00 87.68           C
+ATOM    786  CB  LYS A  97       9.976  -3.216  13.709  1.00 87.68           C
+ATOM    787  O   LYS A  97       9.131  -5.256  16.109  1.00 87.68           O
+ATOM    788  CG  LYS A  97      10.961  -3.264  12.550  1.00 87.68           C
+ATOM    789  CD  LYS A  97      11.875  -2.045  12.542  1.00 87.68           C
+ATOM    790  CE  LYS A  97      12.840  -2.076  11.365  1.00 87.68           C
+ATOM    791  NZ  LYS A  97      13.742  -0.886  11.356  1.00 87.68           N
+ATOM    792  N   GLU A  98       7.395  -3.949  15.535  1.00 85.16           N
+ATOM    793  CA  GLU A  98       6.766  -3.985  16.852  1.00 85.16           C
+ATOM    794  C   GLU A  98       6.185  -5.364  17.149  1.00 85.16           C
+ATOM    795  CB  GLU A  98       5.671  -2.920  16.954  1.00 85.16           C
+ATOM    796  O   GLU A  98       6.174  -5.804  18.300  1.00 85.16           O
+ATOM    797  CG  GLU A  98       6.201  -1.494  16.993  1.00 85.16           C
+ATOM    798  CD  GLU A  98       5.199  -0.496  17.550  1.00 85.16           C
+ATOM    799  OE1 GLU A  98       5.207  -0.247  18.777  1.00 85.16           O
+ATOM    800  OE2 GLU A  98       4.397   0.041  16.753  1.00 85.16           O
+ATOM    801  N   LYS A  99       5.862  -6.201  16.145  1.00 71.59           N
+ATOM    802  CA  LYS A  99       5.236  -7.516  16.249  1.00 71.59           C
+ATOM    803  C   LYS A  99       6.285  -8.620  16.338  1.00 71.59           C
+ATOM    804  CB  LYS A  99       4.311  -7.766  15.057  1.00 71.59           C
+ATOM    805  O   LYS A  99       6.031  -9.682  16.910  1.00 71.59           O
+ATOM    806  CG  LYS A  99       2.886  -7.276  15.266  1.00 71.59           C
+ATOM    807  CD  LYS A  99       1.932  -7.876  14.240  1.00 71.59           C
+ATOM    808  CE  LYS A  99       0.499  -7.420  14.474  1.00 71.59           C
+ATOM    809  NZ  LYS A  99      -0.480  -8.264  13.726  1.00 71.59           N
+ATOM    810  N   GLU A 100       7.534  -8.377  16.016  1.00 69.93           N
+ATOM    811  CA  GLU A 100       8.637  -9.325  16.142  1.00 69.93           C
+ATOM    812  C   GLU A 100       9.233  -9.296  17.546  1.00 69.93           C
+ATOM    813  CB  GLU A 100       9.723  -9.028  15.104  1.00 69.93           C
+ATOM    814  O   GLU A 100       9.551 -10.343  18.113  1.00 69.93           O
+ATOM    815  CG  GLU A 100       9.683  -9.948  13.892  1.00 69.93           C
+ATOM    816  CD  GLU A 100      10.905  -9.814  12.998  1.00 69.93           C
+ATOM    817  OE1 GLU A 100      11.866 -10.601  13.160  1.00 69.93           O
+ATOM    818  OE2 GLU A 100      10.903  -8.915  12.128  1.00 69.93           O
+ENDMDL
+END

test/model-test.cpp

@@ -54,7 +54,7 @@ TEST_CASE("create_nonpoly_1")
 
 	cif::file file;
 	auto &&[dbi, ignore] = file.emplace("TEST"); // create a datablock
-	dbi->load_dictionary("mmcif_pdbx.dic");
+	dbi->set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	cif::mm::structure structure(file);
 
@@ -82,7 +82,7 @@ _atom_site.pdbx_formal_charge
 # that's enough to test with
 )"_cf;
 
-	atoms.front().load_dictionary("mmcif_pdbx.dic");
+	atoms.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	auto &hem_data = atoms["HEM"];
 	auto &atom_site = hem_data["atom_site"];
@@ -159,7 +159,7 @@ _struct_asym.details                       ?
 _atom_type.symbol   C
 )"_cf;
 
-	expected.front().load_dictionary("mmcif_pdbx.dic");
+	expected.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	if (not(expected.front() == structure.get_datablock()))
 	{
@@ -178,7 +178,7 @@ TEST_CASE("create_nonpoly_2")
 
 	cif::file file;
 	auto &&[dbi, ignore] = file.emplace("TEST"); // create a datablock
-	dbi->load_dictionary("mmcif_pdbx.dic");
+	dbi->set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	cif::mm::structure structure(file);
 
@@ -270,7 +270,7 @@ _struct_asym.details                       ?
 _atom_type.symbol   C
 )"_cf;
 
-	expected.front().load_dictionary("mmcif_pdbx.dic");
+	expected.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	REQUIRE(expected.front() == structure.get_datablock());
 
@@ -354,7 +354,7 @@ _struct_asym.details                       ?
 #
 )"_cf;
 
-	data.front().load_dictionary("mmcif_pdbx.dic");
+	data.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
 
 	cif::mm::structure s(data);
 
@@ -431,3 +431,169 @@ TEST_CASE("remove_residue_1")
 
 	REQUIRE_NOTHROW(s.validate_atoms());
 }
+
+// --------------------------------------------------------------------
+// Tests for structure_open_options
+
+TEST_CASE("options_1")
+{
+	using namespace cif::literals;
+
+	const std::filesystem::path example(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	cif::file file(example.string());
+
+	auto &cf = cif::compound_factory::instance();
+
+	SECTION("skip_water")
+	{
+		cif::mm::structure s(file, 1, { .skip_water = true });
+
+		REQUIRE_NOTHROW(s.validate_atoms());
+
+		for (auto a : s.atoms())
+			CHECK_FALSE(a.is_water());
+	}
+
+	SECTION("skip_hetatom")
+	{
+		cif::mm::structure s(file, 1, { .skip_hetatom = true });
+
+		REQUIRE_NOTHROW(s.validate_atoms());
+
+		for (auto a : s.atoms())
+			CHECK((a.is_water() or cf.is_peptide(a.get_label_comp_id()) or cf.is_base(a.get_label_comp_id())));
+	}
+
+	SECTION("selected_asyms")
+	{
+		cif::mm::structure s(file, 1, { .asyms = { "A" } });
+
+		REQUIRE_NOTHROW(s.validate_atoms());
+
+		for (auto a : s.atoms())
+			CHECK(a.get_label_asym_id() == "A");
+	}
+
+	SECTION("min-b-factor")
+	{
+		cif::mm::structure s(file, 1, { .min_b_factor = 20.f });
+
+		REQUIRE_NOTHROW(s.validate_atoms());
+
+		for (auto a : s.atoms())
+			CHECK(a.get_property_float("B_iso_or_equiv") >= 20.f);
+	}
+
+	SECTION("max-b-factor")
+	{
+		cif::mm::structure s(file, 1, { .max_b_factor = 20.f });
+
+		REQUIRE_NOTHROW(s.validate_atoms());
+
+		for (auto a : s.atoms())
+			CHECK(a.get_property_float("B_iso_or_equiv") <= 20.f);
+	}
+}
+
+TEST_CASE("options_2")
+{
+
+	auto data = R"(
+data_TEST
+# 
+_pdbx_nonpoly_scheme.asym_id         A 
+_pdbx_nonpoly_scheme.ndb_seq_num     1 
+_pdbx_nonpoly_scheme.entity_id       1 
+_pdbx_nonpoly_scheme.mon_id          HEM 
+_pdbx_nonpoly_scheme.pdb_seq_num     1 
+_pdbx_nonpoly_scheme.auth_seq_num    1 
+_pdbx_nonpoly_scheme.pdb_mon_id      HEM 
+_pdbx_nonpoly_scheme.auth_mon_id     HEM 
+_pdbx_nonpoly_scheme.pdb_strand_id   A 
+_pdbx_nonpoly_scheme.pdb_ins_code    . 
+#
+loop_
+_atom_site.id
+_atom_site.auth_asym_id
+_atom_site.label_alt_id
+_atom_site.label_asym_id
+_atom_site.label_atom_id
+_atom_site.label_comp_id
+_atom_site.label_entity_id
+_atom_site.label_seq_id
+_atom_site.type_symbol
+_atom_site.group_PDB
+_atom_site.pdbx_PDB_ins_code
+_atom_site.Cartn_x
+_atom_site.Cartn_y
+_atom_site.Cartn_z
+_atom_site.occupancy
+_atom_site.B_iso_or_equiv
+_atom_site.pdbx_formal_charge
+_atom_site.auth_seq_id
+_atom_site.auth_comp_id
+_atom_site.auth_atom_id
+_atom_site.pdbx_PDB_model_num
+1 A A A CHA HEM 1 . C HETATM ? -5.248 39.769 -0.250 0.75 7.67 ? 1 HEM CHA 1
+3 A A A CHB HEM 1 . C HETATM ? -3.774 36.790 3.280  0.75 7.05 ? 1 HEM CHB 1
+2 A A A CHC HEM 1 . C HETATM ? -2.879 33.328 0.013  0.75 7.69 ? 1 HEM CHC 1
+4 A A A CHD HEM 1 . C HETATM ? -4.342 36.262 -3.536 0.75 8.00 ? 1 HEM CHD 1
+5 A B A CHA HEM 1 . C HETATM ? -5.248 39.769 -0.250 0.25 7.67 ? 1 HEM CHA 1
+6 A B A CHB HEM 1 . C HETATM ? -3.774 36.790 3.280  0.25 7.05 ? 1 HEM CHB 1
+7 A B A CHC HEM 1 . C HETATM ? -2.879 33.328 0.013  0.25 7.69 ? 1 HEM CHC 1
+8 A B A CHD HEM 1 . C HETATM ? -4.342 36.262 -3.536 0.25 8.00 ? 1 HEM CHD 1
+#
+_chem_comp.id               HEM
+_chem_comp.type             NON-POLYMER
+_chem_comp.name             'PROTOPORPHYRIN IX CONTAINING FE'
+_chem_comp.formula          'C34 H32 Fe N4 O4'
+_chem_comp.formula_weight   616.487000
+#
+_pdbx_entity_nonpoly.entity_id   1
+_pdbx_entity_nonpoly.name        'PROTOPORPHYRIN IX CONTAINING FE'
+_pdbx_entity_nonpoly.comp_id     HEM
+#
+_entity.id                 1
+_entity.type               non-polymer
+_entity.pdbx_description   'PROTOPORPHYRIN IX CONTAINING FE'
+_entity.formula_weight     616.487000
+#
+_struct_asym.id                            A
+_struct_asym.entity_id                     1
+_struct_asym.pdbx_blank_PDB_chainid_flag   N
+_struct_asym.pdbx_modified                 N
+_struct_asym.details                       ?
+#
+)"_cf;
+
+	data.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+
+	SECTION("max")
+	{
+		cif::mm::structure s(data, 1, {
+			.occupancy_mode = cif::mm::occupancy_policy::MAX
+		});
+
+		REQUIRE(s.atoms().size() == 4);
+		CHECK(s.atoms().front().get_label_alt_id() == "A");
+	}
+
+	SECTION("min")
+	{
+		cif::mm::structure s(data, 1, {
+			.occupancy_mode = cif::mm::occupancy_policy::MIN
+		});
+
+		REQUIRE(s.atoms().size() == 4);
+		CHECK(s.atoms().front().get_label_alt_id() == "B");
+	}
+
+	SECTION("unoccupied")
+	{
+		cif::mm::structure s(data, 1, {
+			.occupancy_mode = cif::mm::occupancy_policy::UNOCCUPIED
+		});
+
+		CHECK(s.atoms().empty());
+	}
+}

test/unit-3d-test.cpp

@@ -90,7 +90,7 @@ TEST_CASE("t1")
 
 	const auto &&[angle, axis] = cif::quaternion_to_angle_axis(q2);
 
-	REQUIRE_THAT(std::fmod(360 + angle, 360), Catch::Matchers::WithinRel(std::fmod(360 - angle0, 360), 0.01));
+	CHECK_THAT(std::fmod(360 + angle, 360), Catch::Matchers::WithinRel(std::fmod(360 - angle0, 360), 0.01));
 
 	for (auto &p : p1)
 		p.rotate(q2);
@@ -116,7 +116,7 @@ TEST_CASE("t2")
 
 	auto &&[angle, axis] = cif::quaternion_to_angle_axis(q);
 
-	REQUIRE_THAT(angle, Catch::Matchers::WithinRel(45.f, 0.01f));
+	CHECK_THAT(angle, Catch::Matchers::WithinRel(45.f, 0.01f));
 }
 
 TEST_CASE("t3")
@@ -140,7 +140,7 @@ TEST_CASE("t3")
 
 	double a = cif::angle(v, p[0], p[1]);
 
-	REQUIRE_THAT(a, Catch::Matchers::WithinRel(45.f, 0.01f));
+	CHECK_THAT(a, Catch::Matchers::WithinRel(45.f, 0.01f));
 }
 
 TEST_CASE("dh_q_0")
@@ -156,7 +156,7 @@ TEST_CASE("dh_q_0")
 	};
 
 	auto a = cif::dihedral_angle(t[0], t[1], t[2], p);
-	REQUIRE_THAT(a, Catch::Matchers::WithinRel(0.f, 0.01f));
+	CHECK_THAT(a, Catch::Matchers::WithinRel(0.f, 0.01f));
 
 	auto q = cif::construct_from_angle_axis(90, axis);
 
@@ -220,7 +220,7 @@ TEST_CASE("dh_q_1")
 		pts[3].rotate(q, pts[2]);
 
 		auto dh = cif::dihedral_angle(pts[0], pts[1], pts[2], pts[3]);
-		REQUIRE_THAT(dh, Catch::Matchers::WithinRel(angle, 0.1f));
+		CHECK_THAT(dh, Catch::Matchers::WithinRel(angle, 0.1f));
 	}
 }
 
@@ -283,9 +283,9 @@ TEST_CASE("dh_q_1")
 
 // 		cif::point p3 = rot * p1;
 
-// 		REQUIRE_THAT(p2.m_x, Catch::Matchers::WithinRel(p3.m_x, 0.01f));
-// 		REQUIRE_THAT(p2.m_y, Catch::Matchers::WithinRel(p3.m_y, 0.01f));
-// 		REQUIRE_THAT(p2.m_z, Catch::Matchers::WithinRel(p3.m_z, 0.01f));
+// 		CHECK_THAT(p2.m_x, Catch::Matchers::WithinRel(p3.m_x, 0.01f));
+// 		CHECK_THAT(p2.m_y, Catch::Matchers::WithinRel(p3.m_y, 0.01f));
+// 		CHECK_THAT(p2.m_z, Catch::Matchers::WithinRel(p3.m_z, 0.01f));
 // 	}
 // }
 
@@ -311,22 +311,22 @@ TEST_CASE("m2q_0a")
 		p2.rotate(q);
 
 		cif::matrix3x3<float> rot_c({
-			rot_c(0, 0) = d[0],
-			rot_c(0, 1) = d[1],
-			rot_c(0, 2) = d[2],
-			rot_c(1, 0) = d[3],
-			rot_c(1, 1) = d[4],
-			rot_c(1, 2) = d[5],
-			rot_c(2, 0) = d[6],
-			rot_c(2, 1) = d[7],
-			rot_c(2, 2) = d[8]
+			static_cast<float>(d[0]),
+			static_cast<float>(d[1]),
+			static_cast<float>(d[2]),
+			static_cast<float>(d[3]),
+			static_cast<float>(d[4]),
+			static_cast<float>(d[5]),
+			static_cast<float>(d[6]),
+			static_cast<float>(d[7]),
+			static_cast<float>(d[8])
 		});
 
 		cif::point p3 = rot_c * p1;
 
-		REQUIRE_THAT(p2.m_x, Catch::Matchers::WithinRel(p3.m_x, 0.01f));
-		REQUIRE_THAT(p2.m_y, Catch::Matchers::WithinRel(p3.m_y, 0.01f));
-		REQUIRE_THAT(p2.m_z, Catch::Matchers::WithinRel(p3.m_z, 0.01f));
+		CHECK_THAT(p2.m_x, Catch::Matchers::WithinRel(p3.m_x, 0.01f));
+		CHECK_THAT(p2.m_y, Catch::Matchers::WithinRel(p3.m_y, 0.01f));
+		CHECK_THAT(p2.m_z, Catch::Matchers::WithinRel(p3.m_z, 0.01f));
 	}
 }
 
@@ -399,15 +399,15 @@ TEST_CASE("symm_1")
 
 	cif::point f = fractional(p, c);
 
-	REQUIRE_THAT(f.m_x, Catch::Matchers::WithinRel(0.1f, 0.01f));
-	REQUIRE_THAT(f.m_y, Catch::Matchers::WithinRel(0.1f, 0.01f));
-	REQUIRE_THAT(f.m_z, Catch::Matchers::WithinRel(0.1f, 0.01f));
+	CHECK_THAT(f.m_x, Catch::Matchers::WithinRel(0.1f, 0.01f));
+	CHECK_THAT(f.m_y, Catch::Matchers::WithinRel(0.1f, 0.01f));
+	CHECK_THAT(f.m_z, Catch::Matchers::WithinRel(0.1f, 0.01f));
 
 	cif::point o = orthogonal(f, c);
 
-	REQUIRE_THAT(o.m_x, Catch::Matchers::WithinRel(1.f, 0.01f));
-	REQUIRE_THAT(o.m_y, Catch::Matchers::WithinRel(1.f, 0.01f));
-	REQUIRE_THAT(o.m_z, Catch::Matchers::WithinRel(1.f, 0.01f));
+	CHECK_THAT(o.m_x, Catch::Matchers::WithinRel(1.f, 0.01f));
+	CHECK_THAT(o.m_y, Catch::Matchers::WithinRel(1.f, 0.01f));
+	CHECK_THAT(o.m_z, Catch::Matchers::WithinRel(1.f, 0.01f));
 }
 
 TEST_CASE("symm_2")
@@ -441,16 +441,16 @@ TEST_CASE("symm_4")
 	cif::point b{ -35.356, 33.693, -3.236 }; // CG2 THR D 400
 	cif::point sb(-6.916, 79.34, 3.236);     // 4_565 copy of b
 
-	REQUIRE_THAT(distance(a, sg(a, c, "1_455"_symop)), Catch::Matchers::WithinRel(static_cast<float>(c.get_a()), 0.01f));
-	REQUIRE_THAT(distance(a, sg(a, c, "1_545"_symop)), Catch::Matchers::WithinRel(static_cast<float>(c.get_b()), 0.01f));
-	REQUIRE_THAT(distance(a, sg(a, c, "1_554"_symop)), Catch::Matchers::WithinRel(static_cast<float>(c.get_c()), 0.01f));
+	CHECK_THAT(distance(a, sg(a, c, "1_455"_symop)), Catch::Matchers::WithinRel(static_cast<float>(c.get_a()), 0.01f));
+	CHECK_THAT(distance(a, sg(a, c, "1_545"_symop)), Catch::Matchers::WithinRel(static_cast<float>(c.get_b()), 0.01f));
+	CHECK_THAT(distance(a, sg(a, c, "1_554"_symop)), Catch::Matchers::WithinRel(static_cast<float>(c.get_c()), 0.01f));
 
 	auto sb2 = sg(b, c, "4_565"_symop);
-	REQUIRE_THAT(sb.m_x, Catch::Matchers::WithinRel(sb2.m_x, 0.01f));
-	REQUIRE_THAT(sb.m_y, Catch::Matchers::WithinRel(sb2.m_y, 0.01f));
-	REQUIRE_THAT(sb.m_z, Catch::Matchers::WithinRel(sb2.m_z, 0.01f));
+	CHECK_THAT(sb.m_x, Catch::Matchers::WithinRel(sb2.m_x, 0.01f));
+	CHECK_THAT(sb.m_y, Catch::Matchers::WithinRel(sb2.m_y, 0.01f));
+	CHECK_THAT(sb.m_z, Catch::Matchers::WithinRel(sb2.m_z, 0.01f));
 
-	REQUIRE_THAT(distance(a, sb2), Catch::Matchers::WithinRel(7.42f, 0.01f));
+	CHECK_THAT(distance(a, sb2), Catch::Matchers::WithinRel(7.42f, 0.01f));
 }
 
 // --------------------------------------------------------------------
@@ -470,17 +470,17 @@ TEST_CASE("symm_4wvp_1")
 	auto a = s.get_residue("A", 10, "").get_atom_by_atom_id("O");
 
 	auto sp1 = c.symmetry_copy(a.get_location(), "2_565"_symop);
-	REQUIRE_THAT(sp1.m_x, Catch::Matchers::WithinAbs(p.m_x, 0.5f));
-	REQUIRE_THAT(sp1.m_y, Catch::Matchers::WithinAbs(p.m_y, 0.5f));
-	REQUIRE_THAT(sp1.m_z, Catch::Matchers::WithinAbs(p.m_z, 0.5f));
+	CHECK_THAT(sp1.m_x, Catch::Matchers::WithinAbs(p.m_x, 0.5f));
+	CHECK_THAT(sp1.m_y, Catch::Matchers::WithinAbs(p.m_y, 0.5f));
+	CHECK_THAT(sp1.m_z, Catch::Matchers::WithinAbs(p.m_z, 0.5f));
 
 	const auto &[d, sp2, so] = c.closest_symmetry_copy(p, a.get_location());
 
 	REQUIRE(d < 1);
 
-	REQUIRE_THAT(sp2.m_x, Catch::Matchers::WithinAbs(p.m_x, 0.5f));
-	REQUIRE_THAT(sp2.m_y, Catch::Matchers::WithinAbs(p.m_y, 0.5f));
-	REQUIRE_THAT(sp2.m_z, Catch::Matchers::WithinAbs(p.m_z, 0.5f));
+	CHECK_THAT(sp2.m_x, Catch::Matchers::WithinAbs(p.m_x, 0.5f));
+	CHECK_THAT(sp2.m_y, Catch::Matchers::WithinAbs(p.m_y, 0.5f));
+	CHECK_THAT(sp2.m_z, Catch::Matchers::WithinAbs(p.m_z, 0.5f));
 }
 
 TEST_CASE("symm_2bi3_1")
@@ -512,17 +512,17 @@ TEST_CASE("symm_2bi3_1")
 		auto sa1 = c.symmetry_copy(a1.get_location(), cif::sym_op(symm1));
 		auto sa2 = c.symmetry_copy(a2.get_location(), cif::sym_op(symm2));
 
-		REQUIRE_THAT(cif::distance(sa1, sa2), Catch::Matchers::WithinAbs(dist, 0.5f));
+		CHECK_THAT(cif::distance(sa1, sa2), Catch::Matchers::WithinAbs(dist, 0.5f));
 
 		auto pa1 = a1.get_location();
 
 		const auto &[d, p, so] = c.closest_symmetry_copy(pa1, a2.get_location());
 
-		REQUIRE_THAT(p.m_x, Catch::Matchers::WithinAbs(sa2.m_x, 0.5f));
-		REQUIRE_THAT(p.m_y, Catch::Matchers::WithinAbs(sa2.m_y, 0.5f));
-		REQUIRE_THAT(p.m_z, Catch::Matchers::WithinAbs(sa2.m_z, 0.5f));
+		CHECK_THAT(p.m_x, Catch::Matchers::WithinAbs(sa2.m_x, 0.5f));
+		CHECK_THAT(p.m_y, Catch::Matchers::WithinAbs(sa2.m_y, 0.5f));
+		CHECK_THAT(p.m_z, Catch::Matchers::WithinAbs(sa2.m_z, 0.5f));
 
-		REQUIRE_THAT(d, Catch::Matchers::WithinAbs(dist, 0.5f));
+		CHECK_THAT(d, Catch::Matchers::WithinAbs(dist, 0.5f));
 		REQUIRE(so.string() == symm2);
 	}
 }
@@ -559,15 +559,15 @@ TEST_CASE("symm_2bi3_1a")
 		auto sa1 = c.symmetry_copy(p1, cif::sym_op(symm1));
 		auto sa2 = c.symmetry_copy(p2, cif::sym_op(symm2));
 
-		REQUIRE_THAT(cif::distance(sa1, sa2), Catch::Matchers::WithinAbs(dist, 0.5f));
+		CHECK_THAT(cif::distance(sa1, sa2), Catch::Matchers::WithinAbs(dist, 0.5f));
 
 		const auto &[d, p, so] = c.closest_symmetry_copy(p1, p2);
 
-		REQUIRE_THAT(p.m_x, Catch::Matchers::WithinAbs(sa2.m_x, 0.5f));
-		REQUIRE_THAT(p.m_y, Catch::Matchers::WithinAbs(sa2.m_y, 0.5f));
-		REQUIRE_THAT(p.m_z, Catch::Matchers::WithinAbs(sa2.m_z, 0.5f));
+		CHECK_THAT(p.m_x, Catch::Matchers::WithinAbs(sa2.m_x, 0.5f));
+		CHECK_THAT(p.m_y, Catch::Matchers::WithinAbs(sa2.m_y, 0.5f));
+		CHECK_THAT(p.m_z, Catch::Matchers::WithinAbs(sa2.m_z, 0.5f));
 
-		REQUIRE_THAT(d, Catch::Matchers::WithinAbs(dist, 0.5f));
+		CHECK_THAT(d, Catch::Matchers::WithinAbs(dist, 0.5f));
 		REQUIRE(so.string() == symm2);
 	}
 }
@@ -590,7 +590,7 @@ TEST_CASE("symm_3bwh_1")
 
 			const auto &[d, p, so] = c.closest_symmetry_copy(a1.get_location(), a2.get_location());
 
-			REQUIRE_THAT(d, Catch::Matchers::WithinAbs(distance(a1.get_location(), p), 0.5f));
+			CHECK_THAT(d, Catch::Matchers::WithinAbs(distance(a1.get_location(), p), 0.5f));
 		}
 	}
 }
@@ -603,5 +603,5 @@ TEST_CASE("volume_3bwh_1")
 
 	cif::crystal c(db);
 
-	REQUIRE_THAT(c.get_cell().get_volume(), Catch::Matchers::WithinRel(741009.625f, 0.01f));
+	CHECK_THAT(c.get_cell().get_volume(), Catch::Matchers::WithinRel(741009.625f, 0.01f));
 }

test/unit-v2-test.cpp

@@ -28,8 +28,6 @@
 
 #include <cif++.hpp>
 
-#include "cif++/dictionary_parser.hpp"
-
 #include <stdexcept>
 
 // --------------------------------------------------------------------
@@ -771,7 +769,7 @@ save__cat_2.desc
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -923,7 +921,7 @@ save__cat_1.c
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -1084,7 +1082,7 @@ save__cat_2.desc
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -1286,7 +1284,7 @@ save__cat_2.parent_id3
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -1506,7 +1504,7 @@ cat_2 3 cat_2:cat_1:3
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -1745,7 +1743,7 @@ cat_2 1 cat_2:cat_1:1
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -2140,7 +2138,7 @@ cat_2 1 '_cat_2.num'  '_cat_3.num'  cat_3
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -2424,7 +2422,7 @@ cat_2 1 '_cat_2.num'  '_cat_3.num'  cat_3
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -3020,7 +3018,7 @@ save__cat_1.name
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -3214,7 +3212,7 @@ save__cat_1.name
 
 	std::istream is_dict(&buffer);
 
-	auto &validator = cif::validator_factory::instance().construct_validator("test_dict.dic", is_dict);
+	auto &validator = cif::validator_factory::instance().add(cif::validator(is_dict));
 
 	cif::file f;
 
@@ -3245,17 +3243,19 @@ _cat_1.name
 	std::istream is_data(&data_buffer);
 	f.load(is_data, validator);
 
-	REQUIRE(f.is_valid());
+	CHECK(f.is_valid());
 
 	std::stringstream ss;
 	ss << f;
 
 	cif::file f2(ss);
-	REQUIRE(f2.is_valid());
+	REQUIRE(f2.empty() == false);
+	f2.front().load_dictionary();
+	CHECK(f2.is_valid());
 
 	auto &audit_conform = f2.front()["audit_conform"];
-	REQUIRE(audit_conform.front()["dict_name"].as<std::string>() == "test_dict.dic");
-	REQUIRE(audit_conform.front()["dict_version"].as<float>() == 1.0);
+	CHECK(audit_conform.front()["dict_name"].as<std::string>() == "test_dict.dic");
+	CHECK(audit_conform.front()["dict_version"].as<float>() == 1.0);
 }
 
 // --------------------------------------------------------------------
@@ -3320,7 +3320,7 @@ save__cat_1.id_2
 
 	std::istream is_dict(&buffer);
 
-	auto validator = cif::parse_dictionary("test", is_dict);
+	cif::validator validator(is_dict);
 
 	cif::file f;
 
@@ -3567,4 +3567,4 @@ TEST_CASE("pdb2cif_formula_weight")
 TEST_CASE("update_values_with_provider")
 {
 
-}
+}

test/validate-pdbx-test.cpp

@@ -279,6 +279,20 @@ A 1 5   GLY 5   5   5   GLY GLY A . n
 
 		REQUIRE_FALSE(cif::pdb::is_valid_pdbx_file(f));
 	}
-
-
+}
+
+TEST_CASE("extended-dictionary-1")
+{
+	cif::add_file_resource("dssp-extension.dic", gTestDir / "dssp-extension.dic");
+
+	cif::VERBOSE = 2;
+
+	auto f = cif::pdb::read(gTestDir / "1cbs-dssp.cif");
+	CHECK(f.is_valid());
+}
+
+TEST_CASE("brak")
+{
+	auto f = cif::pdb::read(gTestDir / "brak.pdb");
+	CHECK(f.is_valid());
 }