rdkit/Code/GraphMol/FragCatalog/FragCatalogEntry.cpp

//
//  Copyright (C) 2003-2022 Greg Landrum and other RDKit contributors
//
//   @@ All Rights Reserved @@
//  This file is part of the RDKit.
//  The contents are covered by the terms of the BSD license
//  which is included in the file license.txt, found at the root
//  of the RDKit source tree.
//
#include "FragCatalogEntry.h"

#include <RDGeneral/types.h>
#include <RDGeneral/utils.h>
#include <RDGeneral/StreamOps.h>
#include <GraphMol/RDKitBase.h>
#include <GraphMol/MolPickler.h>
#include <GraphMol/Subgraphs/SubgraphUtils.h>
#include <GraphMol/Subgraphs/Subgraphs.h>
#include <RDGeneral/hash/hash.hpp>

#include <algorithm>
#include <cstdint>
#include <cstdlib>
#include <fstream>
#include <tuple>

namespace RDKit {

FragCatalogEntry::FragCatalogEntry(const ROMol *omol, const PATH_TYPE &path,
                                   const MatchVectType &aidToFid) {
  PRECONDITION(omol, "bad mol");
  // start with the assumption that this entry is not participating in
  //  any find of fingerprinting
  d_aToFmap.clear();
  setBitId(-1);
  INT_MAP_INT aIdxMap;  // a map from atom id in omol to the new atoms id in mol
  dp_mol = Subgraphs::pathToSubmol(*omol, path, false,
                                   aIdxMap);  // Using Subgraphs functionality
  d_order = path.size();

  // using aIdxMap initialize the location (and their IDs) of the
  // functional groups on dp_mol
  for (const auto &mvtci : aidToFid) {
    int oldAid = mvtci.first;
    if (aIdxMap.find(oldAid) != aIdxMap.end()) {
      int newAid = aIdxMap[oldAid];
      if (d_aToFmap.find(newAid) != d_aToFmap.end()) {
        d_aToFmap[newAid].push_back(mvtci.second);
      } else {
        INT_VECT tmpVect;
        tmpVect.clear();
        tmpVect.push_back(mvtci.second);
        d_aToFmap[newAid] = tmpVect;
      }
    }
  }
  dp_props = new Dict();

  d_descrip = "";
}

FragCatalogEntry::FragCatalogEntry(const std::string &pickle) {
  d_aToFmap.clear();
  dp_props = new Dict();
  this->initFromString(pickle);
}

void FragCatalogEntry::setDescription(const FragCatParams *params) {
  PRECONDITION(params, "");
  for (auto &[atIdx, fGroups] : d_aToFmap) {
    std::string label;
    std::string temp;

    const ROMol *fGroup = nullptr;
    auto fGroupIdx = fGroups.cbegin();
    for (unsigned int i = 0; i < fGroups.size() - 1; i++) {
      fGroup = params->getFuncGroup(*fGroupIdx);
      fGroup->getProp(common_properties::_Name, temp);
      label += "(<" + temp + ">)";
      ++fGroupIdx;
    }
    fGroup = params->getFuncGroup(*fGroupIdx);
    fGroup->getProp(common_properties::_Name, temp);
    label += "<" + temp + ">";
    dp_mol->getAtomWithIdx(atIdx)->setProp(
        common_properties::_supplementalSmilesLabel, label);
  }

  d_descrip = MolToSmiles(*dp_mol);
};

bool FragCatalogEntry::match(const FragCatalogEntry *other, double tol) const {
  PRECONDITION(other, "bad fragment to compare");

  if (d_order != other->getOrder()) {
    return false;
  }
  // now check if both the entries have the same number of functional groups
  const INT_INT_VECT_MAP &oFgpMap = other->getFuncGroupMap();

  if (oFgpMap.size() != d_aToFmap.size()) {
    return false;
  }

  // now check if the IDs are the same
  for (const auto &tfi : d_aToFmap) {
    bool found = false;
    for (const auto &ofi : oFgpMap) {
      if (tfi.second == ofi.second) {
        found = true;
        break;
      }
    }
    if (!found) {
      return false;
    }
  }

  // FIX: if might be better if we just do the balaban first and then
  // move onto eigen values
  Subgraphs::DiscrimTuple tdiscs, odiscs;
  odiscs = other->getDiscrims();

  tdiscs = this->getDiscrims();
  // REVIEW: need an overload of feq that handles tuples in MolOps, or wherever
  // DiscrimTuple is defined
  // FIX: this may not be enough
  // we may have to do the actual isomorphism mapping
  return feq(std::get<0>(tdiscs), std::get<0>(odiscs), tol) &&
         feq(std::get<1>(tdiscs), std::get<1>(odiscs), tol) &&
         feq(std::get<2>(tdiscs), std::get<2>(odiscs), tol);
}

Subgraphs::DiscrimTuple FragCatalogEntry::getDiscrims() const {
  Subgraphs::DiscrimTuple res;
  if (this->hasProp(common_properties::Discrims)) {
    this->getProp(common_properties::Discrims, res);
  } else {
    PATH_TYPE path(dp_mol->getNumBonds(), 0);
    std::iota(path.begin(), path.end(), 0);

    // create invariant additions to reflect the functional groups attached to
    // the atoms
    gboost::hash<INT_VECT> vectHasher;
    std::vector<std::uint32_t> funcGpInvars(dp_mol->getNumAtoms(), 0);
    for (unsigned int aid = 0; aid < dp_mol->getNumAtoms(); ++aid) {
      std::uint32_t invar = 0;
      auto mapPos = d_aToFmap.find(aid);
      if (mapPos != d_aToFmap.end()) {
        INT_VECT fGroups = mapPos->second;
        std::sort(fGroups.begin(), fGroups.end());
        invar = vectHasher(fGroups);
      }
      funcGpInvars[aid] = invar;
    }
    res = Subgraphs::calcPathDiscriminators(*dp_mol, path, true, &funcGpInvars);
    this->setProp(common_properties::Discrims, res);
  }

  // std::cout << "DISCRIMS: " << d_descrip  << " ";
  // std::cout << res.get<0>() << " " << res.get<1>() << " " << res.get<2>();
  // std::cout << std::endl;
  return res;
}

void FragCatalogEntry::toStream(std::ostream &ss) const {
  MolPickler::pickleMol(*dp_mol, ss);

  std::int32_t tmpInt;
  tmpInt = getBitId();
  streamWrite(ss, tmpInt);

  tmpInt = d_descrip.size();
  streamWrite(ss, tmpInt);
  ss.write(d_descrip.c_str(), tmpInt * sizeof(char));

  tmpInt = d_order;
  streamWrite(ss, tmpInt);

  tmpInt = d_aToFmap.size();
  streamWrite(ss, tmpInt);
  for (const auto &iivmci : d_aToFmap) {
    tmpInt = iivmci.first;
    streamWrite(ss, tmpInt);
    INT_VECT tmpVect = iivmci.second;
    tmpInt = tmpVect.size();
    streamWrite(ss, tmpInt);
    for (auto tmpInt : tmpVect) {
      streamWrite(ss, tmpInt);
    }
  }
}

std::string FragCatalogEntry::Serialize() const {
  std::stringstream ss(std::ios_base::binary | std::ios_base::out |
                       std::ios_base::in);
  toStream(ss);
  return ss.str();
}

void FragCatalogEntry::initFromStream(std::istream &ss) {
  // the molecule:
  dp_mol = new ROMol();
  MolPickler::molFromPickle(ss, *dp_mol);

  std::int32_t tmpInt;
  // the bitId:
  streamRead(ss, tmpInt);
  setBitId(tmpInt);

  // the description:
  streamRead(ss, tmpInt);
  auto *tmpText = new char[tmpInt + 1];
  ss.read(tmpText, tmpInt * sizeof(char));
  tmpText[tmpInt] = 0;
  d_descrip = tmpText;
  delete[] tmpText;

  streamRead(ss, tmpInt);
  d_order = tmpInt;

  // now the map:
  streamRead(ss, tmpInt);
  for (int i = 0; i < tmpInt; i++) {
    std::int32_t key, value, size;
    streamRead(ss, key);
    streamRead(ss, size);
    INT_VECT tmpVect;
    tmpVect.clear();
    for (int j = 0; j < size; j++) {
      streamRead(ss, value);
      tmpVect.push_back(value);
    }
    d_aToFmap[key] = tmpVect;
  }
}

void FragCatalogEntry::initFromString(const std::string &text) {
  std::stringstream ss(std::ios_base::binary | std::ios_base::out |
                       std::ios_base::in);
  // initialize the stream:
  ss.write(text.c_str(), text.length());
  // now start reading out values:
  initFromStream(ss);
}
}  // namespace RDKit