void test6() {
BOOST_LOG(rdErrorLog) << "-------------------------------------" << std::endl;
BOOST_LOG(rdErrorLog) << "Feature Location testing." << std::endl;
ROMol *testMol;
Conformer *conf;
std::string inText;
FeatSPtrList featSPtrs;
boost::shared_ptr<MolChemicalFeature> featSPtr;
MolChemicalFeatureFactory *factory;
MolChemicalFeatureDef::CollectionType::value_type featDef;
inText =
"DefineFeature HDonor1 [N,O;!H0]\n"
" Family HBondDonor\n"
" Weights 1.0\n"
"EndFeature\n"
"DefineFeature Carboxyl1 C(=O)[O;H1,-]\n"
" Family ZnBinder\n"
" Weights 1.0,1.0,1.0\n"
"EndFeature\n";
factory = buildFeatureFactory(inText);
TEST_ASSERT(factory);
TEST_ASSERT(factory->getNumFeatureDefs() == 2);
testMol = SmilesToMol("C(=O)O");
TEST_ASSERT(testMol);
conf = new Conformer(3);
testMol->addConformer(conf);
conf->setAtomPos(0, RDGeom::Point3D(0, 0, 0.0));
conf->setAtomPos(1, RDGeom::Point3D(1.2, 0, 0.0));
conf->setAtomPos(2, RDGeom::Point3D(0, 1.5, 0.0));
featSPtrs = factory->getFeaturesForMol(*testMol);
TEST_ASSERT(featSPtrs.size() == 2);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "HBondDonor");
TEST_ASSERT(featSPtr->getType() == "HDonor1");
TEST_ASSERT(feq(featSPtr->getPos().x, 0.0));
TEST_ASSERT(feq(featSPtr->getPos().y, 1.5));
TEST_ASSERT(feq(featSPtr->getPos().z, 0.0));
featSPtr = *(++featSPtrs.begin());
TEST_ASSERT(featSPtr->getFamily() == "ZnBinder");
TEST_ASSERT(featSPtr->getType() == "Carboxyl1");
TEST_ASSERT(feq(featSPtr->getPos().x, 0.4));
TEST_ASSERT(feq(featSPtr->getPos().y, 0.5));
TEST_ASSERT(feq(featSPtr->getPos().z, 0.0));
delete testMol;
delete factory;
BOOST_LOG(rdErrorLog) << " done" << std::endl;
}
FeatSPtr getMolFeature(const MolChemicalFeatureFactory &factory,
const ROMol &mol, int idx, std::string includeOnly = "",
bool recompute = true) {
static FeatSPtrList feats;
if (recompute) {
feats = factory.getFeaturesForMol(mol, includeOnly.c_str());
}
if (idx < 0 || idx >= static_cast<int>(feats.size())) {
throw IndexErrorException(idx);
}
FeatSPtrList_I fi = feats.begin();
for (int i = 0; i < idx; ++i) {
++fi;
}
return (*fi);
}
void testGithub252() {
BOOST_LOG(rdErrorLog) << "-------------------------------------" << std::endl;
BOOST_LOG(rdErrorLog) << "Testing Github Issue #252: crash when calling "
"getPos() with no conformer."
<< std::endl;
BOOST_LOG(rdErrorLog) << " expect a precondition failure message below"
<< std::endl;
ROMol *testMol;
MolChemicalFeatureFactory *factory;
FeatSPtrList featSPtrs;
boost::shared_ptr<MolChemicalFeature> featSPtr;
std::string fdef =
"DefineFeature CTriplet [C][C][C]\n"
" Family LumpedHydrophobe\n"
" Weights 1.0,1.0,1.0\n"
"EndFeature\n"
"DefineFeature CPair [C][C]\n"
" Family LumpedHydrophobe\n"
" Weights 1.0,1.0\n"
"EndFeature\n";
factory = buildFeatureFactory(fdef);
TEST_ASSERT(factory);
TEST_ASSERT(factory->getNumFeatureDefs() == 2);
testMol = SmilesToMol("CCC");
TEST_ASSERT(testMol);
featSPtrs = factory->getFeaturesForMol(*testMol);
TEST_ASSERT(featSPtrs.size() == 1);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "LumpedHydrophobe");
bool ok = false;
try {
featSPtr->getPos();
} catch (const Invar::Invariant &) {
ok = true;
}
delete factory;
delete testMol;
TEST_ASSERT(ok);
BOOST_LOG(rdErrorLog) << " Done" << std::endl;
}
void testIssue347() {
BOOST_LOG(rdErrorLog) << "-------------------------------------" << std::endl;
BOOST_LOG(rdErrorLog) << "Test Issue347" << std::endl;
ROMol *testMol;
MolChemicalFeatureFactory *factory;
MolChemicalFeatureDef::CollectionType::const_iterator featDefIt;
MolChemicalFeatureDef::CollectionType::value_type featDef;
FeatSPtrList featSPtrs;
boost::shared_ptr<MolChemicalFeature> featSPtr;
std::string fdef;
fdef =
"DefineFeature CTriplet [C][C][C]\n"
" Family LumpedHydrophobe\n"
" Weights 1.0,1.0,1.0\n"
"EndFeature\n"
"DefineFeature CPair [C][C]\n"
" Family LumpedHydrophobe\n"
" Weights 1.0,1.0\n"
"EndFeature\n";
factory = buildFeatureFactory(fdef);
TEST_ASSERT(factory);
TEST_ASSERT(factory->getNumFeatureDefs() == 2);
featDefIt = factory->beginFeatureDefs();
featDef = *featDefIt;
TEST_ASSERT(featDef->getFamily() == "LumpedHydrophobe");
featDefIt++;
featDef = *featDefIt;
TEST_ASSERT(featDef->getFamily() == "LumpedHydrophobe");
featDefIt++;
TEST_ASSERT(featDefIt == factory->endFeatureDefs());
testMol = SmilesToMol("CCC");
TEST_ASSERT(testMol);
featSPtrs = factory->getFeaturesForMol(*testMol);
TEST_ASSERT(featSPtrs.size() == 1);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "LumpedHydrophobe");
TEST_ASSERT(featSPtr->getType() == "CTriplet");
delete factory;
// now reverse the order and we should get two matches:
fdef =
"DefineFeature CPair [C][C]\n"
" Family LumpedHydrophobe\n"
" Weights 1.0,1.0\n"
"EndFeature\n"
"DefineFeature CTriplet [C][C][C]\n"
" Family LumpedHydrophobe\n"
" Weights 1.0,1.0,1.0\n"
"EndFeature\n";
factory = buildFeatureFactory(fdef);
TEST_ASSERT(factory);
TEST_ASSERT(factory->getNumFeatureDefs() == 2);
featDefIt = factory->beginFeatureDefs();
featDef = *featDefIt;
TEST_ASSERT(featDef->getFamily() == "LumpedHydrophobe");
featDefIt++;
featDef = *featDefIt;
TEST_ASSERT(featDef->getFamily() == "LumpedHydrophobe");
featDefIt++;
TEST_ASSERT(featDefIt == factory->endFeatureDefs());
featSPtrs = factory->getFeaturesForMol(*testMol);
TEST_ASSERT(featSPtrs.size() == 3);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "LumpedHydrophobe");
TEST_ASSERT(featSPtr->getType() == "CPair");
featSPtr = *(++featSPtrs.begin());
TEST_ASSERT(featSPtr->getFamily() == "LumpedHydrophobe");
TEST_ASSERT(featSPtr->getType() == "CPair");
featSPtr = *(++++featSPtrs.begin());
TEST_ASSERT(featSPtr->getFamily() == "LumpedHydrophobe");
TEST_ASSERT(featSPtr->getType() == "CTriplet");
delete factory;
delete testMol;
BOOST_LOG(rdErrorLog) << " done" << std::endl;
}
void test5() {
BOOST_LOG(rdErrorLog) << "-------------------------------------" << std::endl;
BOOST_LOG(rdErrorLog) << "FeatureFactory testing." << std::endl;
ROMol *testMol;
MatchVectType mv;
std::string inText;
FeatSPtrList featSPtrs;
boost::shared_ptr<MolChemicalFeature> featSPtr;
MolChemicalFeatureFactory *factory;
MolChemicalFeatureDef::CollectionType::value_type featDef;
inText =
"DefineFeature HDonor1 [N,O;!H0]\n"
" Family HBondDonor\n"
" Weights 1.0\n"
"EndFeature\n"
"DefineFeature HAcceptor1 [N,O]\n"
" Family HBondAcceptor\n"
" Weights 1.0\n"
"EndFeature\n";
factory = buildFeatureFactory(inText);
TEST_ASSERT(factory);
TEST_ASSERT(factory->getNumFeatureDefs() == 2);
testMol = SmilesToMol("COCN");
TEST_ASSERT(testMol);
featDef = *factory->beginFeatureDefs();
TEST_ASSERT(SubstructMatch(*testMol, *featDef->getPattern(), mv));
BOOST_LOG(rdErrorLog) << "1" << std::endl;
TEST_ASSERT(mv.size() == 1);
TEST_ASSERT(mv[0].first == 0);
TEST_ASSERT(mv[0].second == 3);
featDef = *(++factory->beginFeatureDefs());
TEST_ASSERT(SubstructMatch(*testMol, *featDef->getPattern(), mv));
BOOST_LOG(rdErrorLog) << "2" << std::endl;
TEST_ASSERT(mv.size() == 1);
TEST_ASSERT(mv[0].first == 0);
TEST_ASSERT(mv[0].second == 1 || mv[0].second == 3);
// Test using the factory to find features:
featSPtrs = factory->getFeaturesForMol(*testMol);
BOOST_LOG(rdErrorLog) << "3" << std::endl;
TEST_ASSERT(featSPtrs.size() == 3);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "HBondDonor");
TEST_ASSERT(featSPtr->getType() == "HDonor1");
featSPtr = *(++featSPtrs.begin());
TEST_ASSERT(featSPtr->getFamily() == "HBondAcceptor");
TEST_ASSERT(featSPtr->getType() == "HAcceptor1");
featSPtr = *(++featSPtrs.begin());
TEST_ASSERT(featSPtr->getFamily() == "HBondAcceptor");
TEST_ASSERT(featSPtr->getType() == "HAcceptor1");
// Test limiting stuff with includeOnly
featSPtrs = factory->getFeaturesForMol(*testMol, "HBondAcceptor");
BOOST_LOG(rdErrorLog) << "4" << std::endl;
TEST_ASSERT(featSPtrs.size() == 2);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "HBondAcceptor");
TEST_ASSERT(featSPtr->getType() == "HAcceptor1");
featSPtr = *(++featSPtrs.begin());
TEST_ASSERT(featSPtr->getFamily() == "HBondAcceptor");
TEST_ASSERT(featSPtr->getType() == "HAcceptor1");
featSPtrs = factory->getFeaturesForMol(*testMol, "HBondDonor");
BOOST_LOG(rdErrorLog) << "5" << std::endl;
TEST_ASSERT(featSPtrs.size() == 1);
featSPtr = *featSPtrs.begin();
TEST_ASSERT(featSPtr->getFamily() == "HBondDonor");
TEST_ASSERT(featSPtr->getType() == "HDonor1");
featSPtrs = factory->getFeaturesForMol(*testMol, "NotPresent");
BOOST_LOG(rdErrorLog) << "6" << std::endl;
TEST_ASSERT(featSPtrs.size() == 0);
delete testMol;
delete factory;
BOOST_LOG(rdErrorLog) << " done" << std::endl;
}
int getNumMolFeatures(const MolChemicalFeatureFactory &factory,
const ROMol &mol, std::string includeOnly = "") {
FeatSPtrList feats = factory.getFeaturesForMol(mol, includeOnly.c_str());
return feats.size();
}
FeatSPtrList MolChemicalFeatureFactory::getFeaturesForMol(const ROMol &mol,
const char* includeOnly) const {
PRECONDITION(includeOnly,"bad limits");
std::string limits(includeOnly);
#ifdef USE_VFLIB
AR_MOLGRAPH *molG=getMolGraph(mol);
#endif
FeatSPtrList res;
int idx = 1;
typedef std::vector< std::pair< std::string,std::set<int> > > MatchSetCollection;
MatchSetCollection matchSets;
for(MolChemicalFeatureDef::CollectionType::const_iterator featDefIt=beginFeatureDefs();
featDefIt!=endFeatureDefs();featDefIt++){
MolChemicalFeatureDef::CollectionType::value_type featDef=*featDefIt;
if(limits=="" || limits==featDef->getFamily()){
std::vector< MatchVectType > matches;
#ifdef USE_VFLIB
unsigned int numMatches=SubstructMatch(molG,*featDef->getPattern(),matches);
#else
unsigned int numMatches=SubstructMatch(mol,*featDef->getPattern(),matches);
#endif
for(unsigned int i=0;i<numMatches;i++){
const MatchVectType &match=matches[i];
std::set<int> matchSet;
for(MatchVectType::const_iterator mIt=match.begin();
mIt!=match.end();
++mIt){
matchSet.insert(mIt->second);
}
// loop over the matches we've already found and see if this one
// is unique:
bool unique=true;
for(MatchSetCollection::const_iterator vsiCI=matchSets.begin();
vsiCI!=matchSets.end();
++vsiCI){
if(vsiCI->first==featDef->getFamily() &&
std::includes(vsiCI->second.begin(),vsiCI->second.end(),
matchSet.begin(),matchSet.end())){
unique=false;
break;
}
}
if(unique){
matchSets.push_back(std::make_pair(featDef->getFamily(),matchSet));
// Set up the feature:
MolChemicalFeature *newFeat=new MolChemicalFeature(&mol,this,featDef.get(),idx++);
MolChemicalFeature::AtomPtrContainer &atoms=newFeat->d_atoms;
atoms.resize(match.size());
// set up the atoms:
for(MatchVectType::const_iterator matchIt=match.begin();
matchIt!=match.end();matchIt++){
int atomIdx=matchIt->second;
int queryIdx=matchIt->first;
atoms[queryIdx]=mol.getAtomWithIdx(atomIdx);
}
// finally, add this to our result:
res.push_back(FeatSPtrList::value_type(newFeat));
}
}
}
}
#ifdef USE_VFLIB
#ifndef CACHE_ARMOLGRAPHS
delete molG;
#endif
#endif
return res;
}