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>
225 lines
9.0 KiB
Python
225 lines
9.0 KiB
Python
import unittest
|
|
import numpy as np
|
|
|
|
from rdkit import Chem
|
|
from rdkit.Chem import rdGaussianShape, rdMolTransforms, rdDistGeom
|
|
from rdkit import RDConfig
|
|
from rdkit.Geometry import Point3D
|
|
|
|
|
|
datadir = RDConfig.RDBaseDir + '/Code/GraphMol/GaussianShape/test_data'
|
|
|
|
|
|
class TestCase(unittest.TestCase):
|
|
|
|
def setUp(self):
|
|
suppl = Chem.SDMolSupplier(datadir + '/test1.sdf')
|
|
self.ref = suppl[0]
|
|
self.probe = suppl[1]
|
|
|
|
def test1_Defaults(self):
|
|
tpl = rdGaussianShape.AlignMol(self.ref, self.probe)
|
|
self.assertAlmostEqual(tpl[0], 0.497, places=3)
|
|
self.assertAlmostEqual(tpl[1], 0.760, places=3)
|
|
self.assertAlmostEqual(tpl[2], 0.233, places=3)
|
|
|
|
|
|
def test2_NoColor(self):
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
ovOpts.optimMode = rdGaussianShape.OptimMode.SHAPE_ONLY
|
|
shpOpts= rdGaussianShape.ShapeInputOptions()
|
|
shpOpts.useColors = False
|
|
tpl = rdGaussianShape.AlignMol(self.ref, self.probe , shpOpts, shpOpts, ovOpts)
|
|
self.assertAlmostEqual(tpl[0], 0.760, places=3)
|
|
self.assertAlmostEqual(tpl[1], 0.760, places=3)
|
|
self.assertAlmostEqual(tpl[2], 0.0, places=3)
|
|
|
|
def test3_FromShape(self):
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
shpOpts= rdGaussianShape.ShapeInputOptions()
|
|
shp = rdGaussianShape.ShapeInput(self.ref, -1, shpOpts, ovOpts)
|
|
self.assertAlmostEqual(shp.ShapeVolume, 591.058, places=3)
|
|
self.assertAlmostEqual(shp.ColorVolume, 31.935, places=3)
|
|
self.assertTrue(type(shp) == rdGaussianShape.ShapeInput)
|
|
tpl = rdGaussianShape.AlignMol(shp, self.probe)
|
|
self.assertAlmostEqual(tpl[0], 0.497, places=3)
|
|
self.assertAlmostEqual(tpl[1], 0.760, places=3)
|
|
self.assertAlmostEqual(tpl[2], 0.233, places=3)
|
|
|
|
mol = shp.ShapeToMol()
|
|
self.assertEqual(Chem.MolToSmiles(mol), "CC(=O)Oc1ccccc1C(=O)O")
|
|
mol = shp.ShapeToMol(True)
|
|
self.assertEqual(Chem.MolToSmiles(mol), "CC(=O)Oc1ccccc1C(=O)O.[Xe].[Xe].[Xe].[Xe].[Xe].[Xe]")
|
|
mol = shp.ShapeToMol(False, False)
|
|
self.assertEqual(Chem.MolToSmiles(mol), "C.C.C.C.C.C.C.C.C.C.C.C.C")
|
|
mol = shp.ShapeToMol(True, False)
|
|
self.assertEqual(Chem.MolToSmiles(mol), "C.C.C.C.C.C.C.C.C.C.C.C.C.[Xe].[Xe].[Xe].[Xe].[Xe].[Xe]")
|
|
|
|
def test4_customFeatures(self):
|
|
m1 = Chem.MolFromSmiles(
|
|
"O=CC=O |(-1.75978,0.148897,0;-0.621382,-0.394324,0;0.624061,0.3656,.1;1.7571,-0.120174,.1)|")
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
opts.customFeatures = [[(1, Point3D(-1.75978, 0.148897, 0), 1.0),
|
|
(2, Point3D(1.7571, -0.120174, 0.1), 1.0)]]
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
shp = rdGaussianShape.ShapeInput(m1, -1, opts, ovOpts)
|
|
self.assertEqual(shp.NumAtoms, 4)
|
|
self.assertEqual(shp.NumFeatures, 2)
|
|
m2 = Chem.Mol(m1)
|
|
opts2 = rdGaussianShape.ShapeInputOptions()
|
|
opts2.customFeatures = [[(2, Point3D(-1.75978, 0.148897, 0), 1.0, [1, 2, 3]),
|
|
(1, Point3D(1.7571, -0.120174, 0.1), 1.0, [4, 5, 6])]]
|
|
shp2 = rdGaussianShape.ShapeInput(m2, -1, opts2, ovOpts)
|
|
tpl = rdGaussianShape.AlignShapes(shp, shp2, ovOpts)
|
|
self.assertAlmostEqual(tpl[0], 0.999, places=3)
|
|
self.assertAlmostEqual(tpl[1], 1.000, places=3)
|
|
self.assertAlmostEqual(tpl[2], 0.998, places=3)
|
|
tf = tpl[3]
|
|
self.assertGreater(0.0, tf[0])
|
|
self.assertEqual(1.0, tf[15])
|
|
|
|
# check the getter:
|
|
cfs = opts2.customFeatures
|
|
self.assertEqual(len(cfs), 1)
|
|
self.assertEqual(len(cfs[0]), 2)
|
|
self.assertEqual(len(cfs[0][0]), 4)
|
|
self.assertEqual(cfs[0][0][0], 2)
|
|
self.assertEqual(cfs[0][1][3][0], 4)
|
|
self.assertEqual(cfs[0][1][3][1], 5)
|
|
self.assertEqual(cfs[0][1][3][2], 6)
|
|
|
|
def test5_customFeatures(self):
|
|
m1 = Chem.MolFromSmiles(
|
|
"O=CC=O |(-1.75978,0.148897,0;-0.621382,-0.394324,0;0.624061,0.3656,.1;1.7571,-0.120174,.1)|")
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
opts.customFeatures = [[(1, Point3D(-1.75978, 0.148897,
|
|
0), 1.0), (2, Point3D(1.7571, -0.120174, 0.1), 1.0)]]
|
|
m2 = Chem.Mol(m1)
|
|
opts2 = rdGaussianShape.ShapeInputOptions()
|
|
opts2.customFeatures = [[(2, Point3D(-1.75978, 0.148897,
|
|
0), 1.0), (1, Point3D(1.7571, -0.120174, 0.1), 1.0)]]
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
tpl = rdGaussianShape.AlignMol(m1, m2, opts, opts2, ovOpts)
|
|
self.assertAlmostEqual(tpl[0], 0.999, places=3)
|
|
self.assertAlmostEqual(tpl[1], 1.000, places=3)
|
|
self.assertAlmostEqual(tpl[2], 0.998, places=3)
|
|
|
|
def test6_FixedScore(self):
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
# Just to make sure it's there and returns a value.
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
tpl = rdGaussianShape.ScoreMol(self.ref, self.ref, opts, opts, ovOpts)
|
|
self.assertAlmostEqual(tpl[0], 1.0, places=3)
|
|
self.assertAlmostEqual(tpl[1], 1.0, places=3)
|
|
self.assertAlmostEqual(tpl[2], 1.0, places=3)
|
|
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
opts.useColors = False
|
|
ovOpts.normalize = False
|
|
shp = rdGaussianShape.ShapeInput(self.ref, -1, opts, ovOpts)
|
|
tpl = rdGaussianShape.ScoreMol(shp, self.probe, opts)
|
|
self.assertAlmostEqual(tpl[0], 0.0, places=3)
|
|
self.assertAlmostEqual(tpl[1], 0.0, places=3)
|
|
self.assertAlmostEqual(tpl[2], 0.0, places=3)
|
|
|
|
opts.useColors = True
|
|
shp1 = rdGaussianShape.ShapeInput(self.probe, -1, opts, ovOpts)
|
|
shp2 = rdGaussianShape.ShapeInput(self.probe, -1, opts, ovOpts)
|
|
tpl = rdGaussianShape.ScoreShape(shp1, shp2, ovOpts)
|
|
self.assertAlmostEqual(tpl[0], 1.0, places=3)
|
|
self.assertAlmostEqual(tpl[1], 1.0, places=3)
|
|
self.assertAlmostEqual(tpl[2], 1.0, places=3)
|
|
self.assertAlmostEqual(tpl[3], 751.0, places=1)
|
|
self.assertAlmostEqual(tpl[4], 42.5, places=1)
|
|
|
|
def test7_customAtomRadii(self):
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
opts.allCarbonRadii = False
|
|
opts.atomRadii = [(4, 1.9)]
|
|
shp = rdGaussianShape.ShapeInput(self.ref, -1, opts, ovOpts)
|
|
self.assertAlmostEqual(shp.ShapeVolume, 559.361, places=3)
|
|
|
|
def test8_atomSubset(self):
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
opts.atomSubset = [4, 5, 6, 7, 8, 9]
|
|
opts.useColors = False
|
|
opts.allCarbonRadii = False
|
|
shp = rdGaussianShape.ShapeInput(self.ref, -1, opts, ovOpts)
|
|
self.assertAlmostEqual(shp.ShapeVolume, 259.144, places=3)
|
|
self.assertEqual(shp.ColorVolume, 0.0)
|
|
|
|
def test9_tversky(self):
|
|
scores = rdGaussianShape.AlignMol(self.ref, self.probe)
|
|
self.assertAlmostEqual(scores[0], 0.497, places=3)
|
|
self.assertAlmostEqual(scores[1], 0.760, places=3)
|
|
self.assertAlmostEqual(scores[2], 0.233, places=3)
|
|
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
ovOpts.simAlpha = 0.95
|
|
ovOpts.simBeta = 0.05
|
|
ref_tversky = rdGaussianShape.AlignMol(self.ref, self.probe, overlayOpts=ovOpts)
|
|
self.assertAlmostEqual(ref_tversky[0], 0.700, places=3)
|
|
self.assertAlmostEqual(ref_tversky[1], 0.968, places=3)
|
|
self.assertAlmostEqual(ref_tversky[2], 0.433, places=3)
|
|
|
|
ovOpts.simAlpha = 0.05
|
|
ovOpts.simBeta = 0.95
|
|
fit_tversky = rdGaussianShape.AlignMol(self.ref, self.probe, overlayOpts=ovOpts)
|
|
self.assertAlmostEqual(fit_tversky[0], 0.557, places=3)
|
|
self.assertAlmostEqual(fit_tversky[1], 0.780, places=3)
|
|
self.assertAlmostEqual(fit_tversky[2], 0.335, places=3)
|
|
|
|
def test10_multipleConformers(self):
|
|
|
|
def loadFile(fileName):
|
|
mol = None
|
|
with open(fileName, "r") as f:
|
|
for line in f:
|
|
line = line.strip()
|
|
if not line:
|
|
continue
|
|
m = Chem.MolFromSmiles(line)
|
|
if mol is None:
|
|
mol = m
|
|
else:
|
|
mol.AddConformer(m.GetConformer())
|
|
return mol
|
|
|
|
esofile = datadir + "/esomeprazole_multi.smi"
|
|
esomeprazole = loadFile(esofile)
|
|
|
|
self.assertEqual(esomeprazole.GetNumConformers(), 10)
|
|
ovOpts = rdGaussianShape.ShapeOverlayOptions()
|
|
opts = rdGaussianShape.ShapeInputOptions()
|
|
# Set the to default values, just to show that they can be set.
|
|
opts.sortShapes = True
|
|
opts.includeDummies = True
|
|
opts.shapePruneThreshold = -1.0
|
|
shapes1 = rdGaussianShape.ShapeInput(esomeprazole, -1, opts, ovOpts)
|
|
self.assertEqual(shapes1.NumShapes, 10)
|
|
|
|
ranfile = datadir + "/ranitidine_multi.smi"
|
|
ranitidine = loadFile(ranfile)
|
|
shapes2 = rdGaussianShape.ShapeInput(ranitidine, -1, opts, ovOpts)
|
|
self.assertEqual(shapes2.NumShapes, 10)
|
|
|
|
bestSim, best1, best2, xform = shapes1.BestSimilarity(shapes2)
|
|
self.assertAlmostEqual(bestSim[0], 0.449, places=3)
|
|
self.assertEqual(best1, 2)
|
|
self.assertEqual(best2, 6)
|
|
self.assertEqual(len(xform), 16)
|
|
|
|
stuff = rdGaussianShape.ScoreMoleculeAllConformers(esomeprazole, ranitidine)
|
|
self.assertEqual(len(stuff), 4)
|
|
self.assertEqual(len(stuff[0]), 10)
|
|
self.assertEqual(len(stuff[0][0]), 10)
|
|
self.assertEqual(stuff[1], 8)
|
|
self.assertEqual(stuff[2], 3)
|
|
self.assertAlmostEqual(stuff[0][8][3], 0.449, places=3)
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|