mirror of
https://github.com/rdkit/rdkit.git
synced 2026-06-03 21:44:30 +08:00
9f551aedbe
* First import of GaussianShape. * Tidying. * Custom features. * Optimise. * Optimise. * Return 3 scores rather than 2 including combo score. * Rename useFeatures to useColors. * Python wrappers. * Python tests. * Take out big test. * Add new start mode, as PubChem does it. * Doh! * Fix MolTransforms eigenvalue return. * Two cycle optimisation, mostly working. * Take out bestSoFar score from SCA. * Take out DTYPE. * Tidy out redundant variables. * Optimisation in 2 parts. * More fiddling in pursuit of speed. * Update Python wrapper. * Tweak. * Atom subsets and different radii. * Fix test. * Revert pubchem_shape's test.cpp. * Serialize ShapeInput. * Trigger build * Remove pointers to std::arrays in ShapeInput. * ShapeInput virtual d'tor. * Precondition - ShapeInput needs a molecule with at least 1 conformer. * Rename ShapeInput::d_centroid to ShapeInput::d_canonTrans. * Fix normalization bugs. * Select start mode using moments of inertia rather than eigenvalues of canonical transformation. * Include color features in moments of inertia. * Smidge faster. * Tversky similarity. * Tidy tests. * Tests working on Linux. * Revert force of right handed axes in MolTransforms::computePrincipalAxesAndMomentsFromGyrationMatrix replacing with a comment in the code. * Response to review. * Sneaky allCarbon bug. * add multithreaded test * Response to review. * Doh! Don't recalculate normalization after every transformation. * Re-instate d_normalizationOK. * Re-name functions for fetching canonical transformations. * Separate alpha from coords. * MultiConf works with single conf extraction. * Extract all conformations. Max and best similarities. * Renames d_currConformer to d_activeShape. * Update shapeToMol. * Update shapeToMol. * Changes from synthon shape searching. * Fix normalization of multiple confs. * Update Python wrappers. * Fix shape merge. * Improve bestSimilarity. * Fix python wrapper. * Pull in changes from SynthonShapeSearch: make pruneShapes public. function to negate Alpha values. * clang-tidy suggestions. * clang-tidy suggestions. * Bug in quaternion gradients - we now have only 3 coordinates. * Tidy tests. * Mac result slightly different. * Multi conformer molecule alignment. * Optionally return raw overlap volumes in score functions. * Python wrappers for raw overlap volumes. * Update Python wrapper ShapeInputOptions. * Tidy for PR. * Extra include file. * Extra library * Tidy forward declarations. * Don't prune if threshold < 0.0. * Windows exporty thing. * Check SMILES on merge of ShapeInputs. * PRECONDITION of SMILES on merge of ShapeInputs. * Response to review - rename some functions. * change how overlapVols is passed add a test for it * API suggestions * Response to review. * Remove debugging writes. * Fix Python wrappers. --------- Co-authored-by: David Cosgrove <david@cozchemix.co.uk> Co-authored-by: greg landrum <greg.landrum@gmail.com>
195 lines
8.8 KiB
C++
195 lines
8.8 KiB
C++
//
|
|
// Copyright (C) 2026 David Cosgrove 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.
|
|
//
|
|
// Original author: David Cosgrove (CozChemIx Limited)
|
|
//
|
|
// This is the interface for the functions to perform shape-based molecule
|
|
// alignments and scoring. It is experimental code and the API and/or
|
|
// results may change in future releases.
|
|
|
|
#ifndef RDKIT_GAUSSIANSHAPE_GUARD
|
|
#define RDKIT_GAUSSIANSHAPE_GUARD
|
|
|
|
#include <RDGeneral/export.h>
|
|
#include <Geometry/Transform3D.h>
|
|
#include <GraphMol/GaussianShape/ShapeInput.h>
|
|
#include <GraphMol/GaussianShape/ShapeOverlayOptions.h>
|
|
|
|
namespace RDKit {
|
|
class ROMol;
|
|
class Conformer;
|
|
|
|
namespace GaussianShape {
|
|
|
|
//! Align a shape onto a reference shape.
|
|
/*!
|
|
\param refShape the reference shape
|
|
\param fitShape the shape to align
|
|
\param xform if passed in as non-null, will be populated with the
|
|
transformation matrix that aligns fit onto ref.
|
|
\param overlayOpts options for the overlay
|
|
|
|
\return an array of the combination score of the shape Tversky value and the
|
|
color Tversky value (zero if colors not used) and the individual values. If
|
|
using color features, defaults to RDKit pharmacophore types for the features.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT std::array<double, 3> AlignShape(
|
|
const ShapeInput &refShape, ShapeInput &fitShape,
|
|
RDGeom::Transform3D *xform = nullptr,
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions());
|
|
|
|
//! Align a molecule to a reference shape
|
|
/*!
|
|
\param refShape the reference shape
|
|
\param fit the molecule to align
|
|
\param fitOpts the options for creating the fit shape
|
|
\param xform if passed in as non-null, will be populated with the
|
|
transformation matrix that aligns fit onto ref.
|
|
\param overlayOpts options for setting up and running the overlay
|
|
\param fitConfId (optional) the conformer to use for the fit
|
|
molecule
|
|
|
|
\return an array of the combination score of the shape Tversky value and the
|
|
color Tversky value (zero if colors not used) and the individual values. If
|
|
using color features, defaults to RDKit pharmacophore types for the features.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT std::array<double, 3> AlignMolecule(
|
|
const ShapeInput &refShape, ROMol &fit,
|
|
const ShapeInputOptions &fitOpts = ShapeInputOptions(),
|
|
RDGeom::Transform3D *xform = nullptr,
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions(),
|
|
int fitConfId = -1);
|
|
|
|
//! Align a molecule to a reference molecule
|
|
/*!
|
|
\param ref the reference molecule
|
|
\param fit the molecule to align
|
|
\param refOpts the options for creating the ref shape
|
|
\param fitOpts the options for creating the fit shape
|
|
\param xform if passed in as non-null, will be populated with the
|
|
transformation matrix that aligns fit onto ref.
|
|
\param overlayOpts options for setting up and running the overlay
|
|
\param refConfId (optional) the conformer to use for the reference
|
|
molecule
|
|
\param fitConfId (optional) the conformer to use for the fit
|
|
molecule
|
|
|
|
\return an array of the combination score of the shape Tversky value and the
|
|
color Tversky value (zero if colors not used) and the individual values. If
|
|
using color features, defaults to RDKit pharmacophore types for the features.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT std::array<double, 3> AlignMolecule(
|
|
const ROMol &ref, ROMol &fit,
|
|
const ShapeInputOptions &refOpts = ShapeInputOptions(),
|
|
const ShapeInputOptions &fitOpts = ShapeInputOptions(),
|
|
RDGeom::Transform3D *xform = nullptr,
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions(),
|
|
int refConfId = -1, int fitConfId = -1);
|
|
|
|
//! Calculate scores for the alignment of all conformers of one molecule
|
|
//! onto another. Returns a matrix of the combination scores, the conformer
|
|
//! numbers of the two molecules that gave the best overlay and the
|
|
//! transformation matrix for that overlay if requested. The molecules
|
|
//! themselves are not altered. scores[0][1] is the score of aligning
|
|
//! fit conformer 1 onto ref conformer 0
|
|
/*!
|
|
\param ref the reference molecule
|
|
\param fit the molecule to align
|
|
\param refConfId returns the reference conformer for the best scoring
|
|
overlay
|
|
\param fitConfId returns the fit conformer for the best scoring overlay
|
|
\param combScores the scores for all the overlays. Will be returned sized
|
|
by the number of conformers of the ref and fit molecules.
|
|
combScores[i][j] will be the score for the jth fit
|
|
conformer onto the ith ref conformer.
|
|
\param refOpts the options for creating the ref shape
|
|
\param fitOpts the options for creating the fit shape
|
|
\param overlayOpts options for setting up and running the overlay
|
|
\param xform if passed in as non-null, will be populated with the
|
|
transformation matrix that gives the best-scoring
|
|
overlay.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT void ScoreMoleculeAllConformers(
|
|
const ROMol &ref, const ROMol &fit, int &refConfId, int &fitConfId,
|
|
std::vector<std::vector<double>> &combScores,
|
|
const ShapeInputOptions &refOpts = ShapeInputOptions(),
|
|
const ShapeInputOptions &fitOpts = ShapeInputOptions(),
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions(),
|
|
RDGeom::Transform3D *xform = nullptr);
|
|
|
|
//! Score the overlap of a shape to a reference shape without moving
|
|
// either.
|
|
/*!
|
|
\param refShape the reference shape
|
|
\param fitShape the shape to score
|
|
\param overlayOpts options for controlling the volume calculation
|
|
\param overlapVols if not-null, is filled with the raw overlap volumes
|
|
|
|
\return an array of the combination score of the shape Tversky value and the
|
|
color Tversky value (zero if colors not used) and the individual values. If
|
|
using color features, defaults to RDKit pharmacophore types for the features.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT std::array<double, 3> ScoreShape(
|
|
const ShapeInput &refShape, const ShapeInput &fitShape,
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions(),
|
|
std::pair<double, double> *overlapVols = nullptr);
|
|
|
|
//! Score the overlap of a molecule to a reference shape without moving
|
|
// either.
|
|
/*!
|
|
\param refShape the reference shape
|
|
\param fit the molecule to score
|
|
\param fitOpts the options for creating the fit shape
|
|
\param overlayOpts options for controlling the volume calculation
|
|
\param fitConfId (optional) the conformer to use for the fit
|
|
molecule
|
|
\param overlapVols if not-null, is filled with the raw overlap volumes
|
|
|
|
\return an array of the combination score of the shape Tversky value and the
|
|
color Tversky value (zero if colors not used) and the individual values. If
|
|
using color features, defaults to RDKit pharmacophore types for the features.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT std::array<double, 3> ScoreMolecule(
|
|
const ShapeInput &refShape, const ROMol &fit,
|
|
const ShapeInputOptions &fitOpts = ShapeInputOptions(),
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions(),
|
|
int fitConfId = -1,
|
|
std::pair<double, double> *overlapVols = nullptr);
|
|
|
|
//! Score the overlap of a molecule to a reference molecule without moving
|
|
// either.
|
|
/*!
|
|
\param ref the reference molecule
|
|
\param fit the molecule to score
|
|
\param refOpts the options for creating the ref shape
|
|
\param fitOpts the options for creating the fit shape
|
|
\param overlayOpts options for controlling the volume calculation
|
|
\param refConfId (optional) the conformer to use for the reference
|
|
molecule
|
|
\param fitConfId (optional) the conformer to use for the fit
|
|
molecule
|
|
\param overlapVols if not-null, is filled with the raw overlap volumes
|
|
|
|
\return an array of the combination score of the shape Tverksy value and the
|
|
color Tversky value (zero if colors not used) and the individual values. If
|
|
using color features, defaults to RDKit pharmacophore types for the features.
|
|
*/
|
|
RDKIT_GAUSSIANSHAPE_EXPORT std::array<double, 3> ScoreMolecule(
|
|
const ROMol &ref, const ROMol &fit,
|
|
const ShapeInputOptions &refOpts = ShapeInputOptions(),
|
|
const ShapeInputOptions &fitOpts = ShapeInputOptions(),
|
|
const ShapeOverlayOptions &overlayOpts = ShapeOverlayOptions(),
|
|
int refConfId = -1, int fitConfId = -1,
|
|
std::pair<double, double> *overlapVols = nullptr);
|
|
|
|
} // namespace GaussianShape
|
|
} // namespace RDKit
|
|
|
|
#endif // RDKIT_GAUSSIANSHAPE_GUARD
|