double getAlignmentTransform(const ROMol &prbMol, const ROMol &refMol,
RDGeom::Transform3D &trans,
int prbCid, int refCid,
const MatchVectType *atomMap,
const RDNumeric::DoubleVector *weights, bool reflect,
unsigned int maxIterations) {
RDGeom::Point3DConstPtrVect refPoints, prbPoints;
const Conformer &prbCnf = prbMol.getConformer(prbCid);
const Conformer &refCnf = refMol.getConformer(refCid);
if (atomMap == 0) {
// we have to figure out the mapping between the two molecule
MatchVectType match;
if (SubstructMatch(refMol, prbMol, match)) {
MatchVectType::const_iterator mi;
for (mi = match.begin(); mi != match.end(); mi++) {
prbPoints.push_back(&prbCnf.getAtomPos(mi->first));
refPoints.push_back(&refCnf.getAtomPos(mi->second));
}
} else {
throw MolAlignException("No sub-structure match found between the probe and query mol");
}
} else {
MatchVectType::const_iterator mi;
for (mi = atomMap->begin(); mi != atomMap->end(); mi++) {
prbPoints.push_back(&prbCnf.getAtomPos(mi->first));
refPoints.push_back(&refCnf.getAtomPos(mi->second));
}
}
double ssr = RDNumeric::Alignments::AlignPoints(refPoints, prbPoints,
trans, weights, reflect, maxIterations);
ssr /= (prbPoints.size());
return sqrt(ssr);
}
//*************************************************************************************
//
// Adds 2D coordinates to a molecule using the Depict.dll
//
// ARGUMENTS:
// mol: the molecule to be altered
// tempFilename: (OPTIONAL) the name of the temporary file
//
// RETURNS:
// 1 on success, 0 otherwise
//
// Here's the process by which this works (it's kind of contorted):
// 1) convert the mol to SMILES
// 2) use the DLL to convert the SMILES to a mol file (on disk)
// 3) parse the mol file into a temporary molecule
// 4) do a substructure match from the old molecule to the
// temp one (which may have a different atom numbering or additional
// atoms added).
// 5) Update the positions of the atoms on the old molecule.
// 6) Free the temp molecule.
//
// NOTES:
// - *FIX:* at the moment we're not doing anything to clear up the
// temp file created in this process. It'll always have the same
// name unless the user explicitly asks that we do something different.
// - To use the DLL, it's essential that the COMBICHEM_ROOT and COMBICHEM_RELEASE
// environment variables be set. If this isn't done, this whole process
// will fail.
// - See the notes above about failures when opening the DLL.
//
//*************************************************************************************
int Add2DCoordsToMolDLL(ROMol &mol,std::string tempFilename){
std::string smi=MolToSmiles(mol,true);
int tmp = SmilesToMolFileDLL(smi,tempFilename);
int res = -1;
if(tmp){
// build another mol from that mol file:
RWMol *tmpMol = MolFileToMol(tempFilename,false);
// match it up with the starting mol:
// (We need to do this because the depict.dll conversion
// to a mol file may have added Hs)
MatchVectType matchVect;
bool hasMatch=SubstructMatch(tmpMol,&mol,matchVect);
if(hasMatch){
const Conformer &conf = tmpMol->getCoformer(0);
Coformer *nconf = new Coformer(mol.getNumAtoms());
for(MatchVectType::const_iterator mvi=matchVect.begin();
mvi!=matchVect.end();mvi++){
nconf->setAtomPos(conf.getAtomPos(mvi->first));
}
confId = (int)mol.addConformer(nconf, true);
}
delete tmpMol;
}
return res;
}
bool StructCheckTautomer::applyTautomer(unsigned it) {
if (Options.FromTautomer.size() <= it || Options.ToTautomer.size() <= it) {
if (Options.Verbose)
BOOST_LOG(rdInfoLog) << "ERROR: incorrect Tautomer index it=" << it
<< "\n";
return false;
}
const ROMol &fromTautomer = *Options.FromTautomer[it];
const ROMol &toTautomer = *Options.ToTautomer[it];
if (toTautomer.getNumAtoms() != fromTautomer.getNumAtoms()) {
if (Options.Verbose)
BOOST_LOG(rdInfoLog) << "ERROR: incorrect data toTautomer.getNumAtoms() "
"!= fromTautomer.getNumAtoms()\n";
// incorrect data
// throw(.....);
return false;
}
const unsigned nta = toTautomer.getNumAtoms();
const unsigned ntb = toTautomer.getNumBonds();
MatchVectType match; // The format is (queryAtomIdx, molAtomIdx)
std::vector<unsigned> atomIdxMap(
Mol.getNumAtoms()); // matched tau atom indeces
if (!SubstructMatch(Mol, *Options.FromTautomer[it],
match)) // SSMatch(mp, from_tautomer, SINGLE_MATCH);
return false;
if (Options.Verbose)
BOOST_LOG(rdInfoLog) << "found match for from_tautomer with " << nta
<< " atoms\n";
// init
for (unsigned i = 0; i < Mol.getNumAtoms(); i++) atomIdxMap[i] = -1;
for (MatchVectType::const_iterator mit = match.begin(); mit != match.end();
mit++) {
unsigned tai = mit->first; // From and To Tautomer Atom index
unsigned mai = mit->second; // Mol Atom index
atomIdxMap[mai] = tai;
}
// scan for completely mapped bonds and replace bond order with mapped bond
// from to_tautomer
for (RDKit::BondIterator_ bond = Mol.beginBonds(); bond != Mol.endBonds();
bond++) {
unsigned ti = atomIdxMap[(*bond)->getBeginAtomIdx()];
unsigned tj = atomIdxMap[(*bond)->getEndAtomIdx()];
if (-1 == ti || -1 == tj) continue;
const Bond *tb = toTautomer.getBondBetweenAtoms(ti, tj);
if (tb && (*bond)->getBondType() != tb->getBondType()) {
(*bond)->setBondType(tb->getBondType());
}
}
// apply charge/radical fixes if any
for (unsigned i = 0; i < match.size(); i++) {
Atom &atom = *Mol.getAtomWithIdx(match[i].second);
const Atom &ta = *toTautomer.getAtomWithIdx(match[i].first);
atom.setFormalCharge(ta.getFormalCharge());
atom.setNumRadicalElectrons(ta.getNumRadicalElectrons());
}
return true;
}
void getCrippenAtomContribs(const ROMol &mol,
std::vector< double > &logpContribs,
std::vector< double > &mrContribs,
bool force,
std::vector<unsigned int> *atomTypes,
std::vector<std::string> *atomTypeLabels
){
PRECONDITION(logpContribs.size()==mol.getNumAtoms() &&
mrContribs.size()==mol.getNumAtoms(),
"bad result vector size");
PRECONDITION((!atomTypes || atomTypes->size()==mol.getNumAtoms()),
"bad atomTypes vector");
PRECONDITION((!atomTypeLabels || atomTypeLabels->size()==mol.getNumAtoms()),
"bad atomTypeLabels vector");
if(!force && mol.hasProp("_crippenLogPContribs")){
std::vector<double> tmpVect1,tmpVect2;
mol.getProp("_crippenLogPContribs",tmpVect1);
mol.getProp("_crippenMRContribs",tmpVect2);
if(tmpVect1.size()==mol.getNumAtoms() &&
tmpVect2.size()==mol.getNumAtoms() ){
logpContribs=tmpVect1;
mrContribs=tmpVect2;
return;
}
}
boost::dynamic_bitset<> atomNeeded(mol.getNumAtoms());
atomNeeded.set();
const CrippenParamCollection *params=CrippenParamCollection::getParams();
for(CrippenParamCollection::ParamsVect::const_iterator it=params->begin();
it!=params->end(); ++it){
std::vector<MatchVectType> matches;
SubstructMatch(mol,*(it->dp_pattern.get()),matches,
false,true);
for(std::vector<MatchVectType>::const_iterator matchIt=matches.begin();
matchIt!=matches.end();++matchIt){
int idx=(*matchIt)[0].second;
if(atomNeeded[idx]){
atomNeeded[idx]=0;
logpContribs[idx] = it->logp;
mrContribs[idx] = it->mr;
if(atomTypes) (*atomTypes)[idx]=it->idx;
if(atomTypeLabels) (*atomTypeLabels)[idx]=it->label;
}
}
// no need to keep matching stuff if we already found all the atoms:
if(atomNeeded.none()) break;
}
mol.setProp("_crippenLogPContribs",logpContribs,true);
mol.setProp("_crippenMRContribs",mrContribs,true);
}
bool SmartsMatcher::hasMatch(const ROMol &mol) const {
PRECONDITION(d_pattern.get(), "bad on pattern");
if (d_min_count == 1 && d_max_count == UINT_MAX) {
RDKit::MatchVectType matches;
return SubstructMatch(mol, *d_pattern.get(), matches);
} else { // need to count
const bool uniquify = true;
std::vector<RDKit::MatchVectType> matches;
unsigned int count =
RDKit::SubstructMatch(mol, *d_pattern.get(), matches, uniquify);
return (count >= d_min_count &&
(d_max_count == UINT_MAX || count <= d_max_count));
}
}
void addTorsions(const ROMol &mol, MMFFMolProperties *mmffMolProperties,
ForceFields::ForceField *field,
const std::string &torsionBondSmarts) {
PRECONDITION(field, "bad ForceField");
PRECONDITION(mmffMolProperties, "bad MMFFMolProperties");
PRECONDITION(mmffMolProperties->isValid(),
"missing atom types - invalid force-field");
std::ostream &oStream = mmffMolProperties->getMMFFOStream();
ROMol::ADJ_ITER nbr1Idx;
ROMol::ADJ_ITER end1Nbrs;
ROMol::ADJ_ITER nbr2Idx;
ROMol::ADJ_ITER end2Nbrs;
double totalTorsionEnergy = 0.0;
RDGeom::PointPtrVect points;
if (mmffMolProperties->getMMFFVerbosity()) {
if (mmffMolProperties->getMMFFVerbosity() == MMFF_VERBOSITY_HIGH) {
oStream << "\n"
"T O R S I O N A L\n\n"
"--------------ATOMS--------------- ---ATOM TYPES--- "
"FF TORSION -----FORCE Params-----\n"
" I J K L I J K L "
"CLASS ANGLE ENERGY V1 V2 V3\n"
"-------------------------------------------------------------"
"-----------------------------------------------------"
<< std::endl;
}
points = field->positions();
}
std::vector<MatchVectType> matchVect;
const ROMol *defaultQuery = DefaultTorsionBondSmarts::query();
const ROMol *query = (torsionBondSmarts == DefaultTorsionBondSmarts::string())
? defaultQuery : SmartsToMol(torsionBondSmarts);
TEST_ASSERT(query);
unsigned int nHits = SubstructMatch(mol, *query, matchVect);
if (query != defaultQuery) delete query;
for (unsigned int i = 0; i < nHits; ++i) {
MatchVectType match = matchVect[i];
TEST_ASSERT(match.size() == 2);
int idx2 = match[0].second;
int idx3 = match[1].second;
const Bond *bond = mol.getBondBetweenAtoms(idx2, idx3);
TEST_ASSERT(bond);
const Atom *jAtom = mol.getAtomWithIdx(idx2);
const Atom *kAtom = mol.getAtomWithIdx(idx3);
if (((jAtom->getHybridization() == Atom::SP2) ||
(jAtom->getHybridization() == Atom::SP3)) &&
((kAtom->getHybridization() == Atom::SP2) ||
(kAtom->getHybridization() == Atom::SP3))) {
ROMol::OEDGE_ITER beg1, end1;
boost::tie(beg1, end1) = mol.getAtomBonds(jAtom);
while (beg1 != end1) {
const Bond *tBond1 = mol[*beg1].get();
if (tBond1 != bond) {
int idx1 = tBond1->getOtherAtomIdx(idx2);
ROMol::OEDGE_ITER beg2, end2;
boost::tie(beg2, end2) = mol.getAtomBonds(kAtom);
while (beg2 != end2) {
const Bond *tBond2 = mol[*beg2].get();
if ((tBond2 != bond) && (tBond2 != tBond1)) {
int idx4 = tBond2->getOtherAtomIdx(idx3);
// make sure this isn't a three-membered ring:
if (idx4 != idx1) {
// we now have a torsion involving atoms (bonds):
// bIdx - (tBond1) - idx1 - (bond) - idx2 - (tBond2) - eIdx
unsigned int torType;
MMFFTor mmffTorParams;
if (mmffMolProperties->getMMFFTorsionParams(
mol, idx1, idx2, idx3, idx4, torType, mmffTorParams)) {
TorsionAngleContrib *contrib = new TorsionAngleContrib(
field, idx1, idx2, idx3, idx4, &mmffTorParams);
field->contribs().push_back(ForceFields::ContribPtr(contrib));
if (mmffMolProperties->getMMFFVerbosity()) {
const Atom *iAtom = mol.getAtomWithIdx(idx1);
const Atom *lAtom = mol.getAtomWithIdx(idx4);
unsigned int iAtomType =
mmffMolProperties->getMMFFAtomType(idx1);
unsigned int jAtomType =
mmffMolProperties->getMMFFAtomType(idx2);
unsigned int kAtomType =
mmffMolProperties->getMMFFAtomType(idx3);
unsigned int lAtomType =
mmffMolProperties->getMMFFAtomType(idx4);
const RDGeom::Point3D p1((*(points[idx1]))[0],
(*(points[idx1]))[1],
(*(points[idx1]))[2]);
const RDGeom::Point3D p2((*(points[idx2]))[0],
(*(points[idx2]))[1],
(*(points[idx2]))[2]);
const RDGeom::Point3D p3((*(points[idx3]))[0],
(*(points[idx3]))[1],
(*(points[idx3]))[2]);
const RDGeom::Point3D p4((*(points[idx4]))[0],
(*(points[idx4]))[1],
(*(points[idx4]))[2]);
const double cosPhi =
MMFF::Utils::calcTorsionCosPhi(p1, p2, p3, p4);
const double torsionEnergy = MMFF::Utils::calcTorsionEnergy(
mmffTorParams.V1, mmffTorParams.V2, mmffTorParams.V3,
cosPhi);
if (mmffMolProperties->getMMFFVerbosity() ==
MMFF_VERBOSITY_HIGH) {
oStream
<< std::left << std::setw(2) << iAtom->getSymbol()
<< " #" << std::setw(5) << idx1 + 1 << std::setw(2)
<< jAtom->getSymbol() << " #" << std::setw(5)
<< idx2 + 1 << std::setw(2) << kAtom->getSymbol()
<< " #" << std::setw(5) << idx3 + 1 << std::setw(2)
<< lAtom->getSymbol() << " #" << std::setw(5)
<< idx4 + 1 << std::right << std::setw(5) << iAtomType
<< std::setw(5) << jAtomType << std::setw(5)
<< kAtomType << std::setw(5) << lAtomType
<< std::setw(6) << torType << " " << std::fixed
<< std::setprecision(3) << std::setw(10)
<< RAD2DEG * acos(cosPhi) << std::setw(10)
<< torsionEnergy << std::setw(10) << mmffTorParams.V1
<< std::setw(10) << mmffTorParams.V2 << std::setw(10)
<< mmffTorParams.V3 << std::endl;
}
totalTorsionEnergy += torsionEnergy;
}
}
}
}
beg2++;
}
}
beg1++;
}
}
}
if (mmffMolProperties->getMMFFVerbosity()) {
if (mmffMolProperties->getMMFFVerbosity() == MMFF_VERBOSITY_HIGH) {
oStream << std::endl;
}
oStream << "TOTAL TORSIONAL ENERGY =" << std::right << std::setw(16)
<< std::fixed << std::setprecision(4) << totalTorsionEnergy
<< std::endl;
}
}
// ------------------------------------------------------------------------
//
//
//
// ------------------------------------------------------------------------
void addTorsions(const ROMol &mol,const AtomicParamVect ¶ms,
ForceFields::ForceField *field,
std::string torsionBondSmarts){
PRECONDITION(mol.getNumAtoms()==params.size(),"bad parameters");
PRECONDITION(field,"bad forcefield");
// find all of the torsion bonds:
std::vector<MatchVectType> matchVect;
ROMol *query=SmartsToMol(torsionBondSmarts);
TEST_ASSERT(query);
unsigned int nHits=SubstructMatch(mol,*query,matchVect);
delete query;
for(unsigned int i=0; i<nHits; i++){
MatchVectType match=matchVect[i];
TEST_ASSERT(match.size()==2);
int idx1=match[0].second;
int idx2=match[1].second;
if(!params[idx1]||!params[idx2]) continue;
const Bond *bond=mol.getBondBetweenAtoms(idx1,idx2);
std::vector<TorsionAngleContrib *> contribsHere;
TEST_ASSERT(bond);
const Atom *atom1=mol.getAtomWithIdx(idx1);
const Atom *atom2=mol.getAtomWithIdx(idx2);
if( (atom1->getHybridization()==Atom::SP2||atom1->getHybridization()==Atom::SP3) &&
(atom2->getHybridization()==Atom::SP2||atom2->getHybridization()==Atom::SP3) ){
ROMol::OEDGE_ITER beg1,end1;
boost::tie(beg1,end1) = mol.getAtomBonds(atom1);
while(beg1!=end1){
const Bond *tBond1=mol[*beg1].get();
if(tBond1!=bond){
int bIdx = tBond1->getOtherAtomIdx(idx1);
ROMol::OEDGE_ITER beg2,end2;
boost::tie(beg2,end2) = mol.getAtomBonds(atom2);
while(beg2 != end2){
const Bond *tBond2=mol[*beg2].get();
if(tBond2!=bond && tBond2!=tBond1){
int eIdx=tBond2->getOtherAtomIdx(idx2);
// make sure this isn't a three-membered ring:
if(eIdx != bIdx){
// we now have a torsion involving atoms (bonds):
// bIdx - (tBond1) - idx1 - (bond) - idx2 - (tBond2) - eIdx
TorsionAngleContrib *contrib;
// if either of the end atoms is SP2 hybridized, set a flag
// here.
bool hasSP2=false;
if(mol.getAtomWithIdx(bIdx)->getHybridization()==Atom::SP2 ||
mol.getAtomWithIdx(bIdx)->getHybridization()==Atom::SP2) {
hasSP2 = true;
}
//std::cout << "Torsion: " << bIdx << "-" << idx1 << "-" << idx2 << "-" << eIdx << std::endl;
//if(okToIncludeTorsion(mol,bond,bIdx,idx1,idx2,eIdx)){
//std::cout << " INCLUDED" << std::endl;
contrib = new TorsionAngleContrib(field,bIdx,idx1,idx2,eIdx,
bond->getBondTypeAsDouble(),
atom1->getAtomicNum(),
atom2->getAtomicNum(),
atom1->getHybridization(),
atom2->getHybridization(),
params[idx1],params[idx2],
hasSP2);
field->contribs().push_back(ForceFields::ContribPtr(contrib));
contribsHere.push_back(contrib);
//}
}
}
beg2++;
}
}
beg1++;
}
}
// now divide the force constant for each contribution to the torsion energy
// about this bond by the number of contribs about this bond:
for(std::vector<TorsionAngleContrib *>::iterator chI=contribsHere.begin();
chI!=contribsHere.end();++chI){
(*chI)->scaleForceConstant(contribsHere.size());
}
}
}
MatchVectType findFuncGroupsOnMol(const ROMol &mol,
const FragCatParams *params,
INT_VECT &fgBonds) {
PRECONDITION(params,"bad params");
fgBonds.clear();
std::pair<int, int> amat;
MatchVectType aidFgrps;
std::vector<MatchVectType> fgpMatches;
std::vector<MatchVectType>::const_iterator mati;
MatchVectType::const_iterator mi;
int aid;
//const ROMol *fgrp;
INT_VECT_CI bi;
aidFgrps.clear();
int fid = 0;
const MOL_SPTR_VECT &fgrps = params->getFuncGroups();
MOL_SPTR_VECT::const_iterator fgci;
for (fgci = fgrps.begin(); fgci != fgrps.end(); fgci++) {
const ROMol *fgrp = fgci->get();
std::string fname;
(*fgci)->getProp(common_properties::_Name, fname);
//std::cout << "Groups number: " << fname << "\n";
//(*fgci)->debugMol(std::cout);
//mol->debugMol(std::cout);
// match this functional group onto the molecule
SubstructMatch(mol, *fgrp, fgpMatches);
// loop over all the matches we get for this fgroup
for (mati = fgpMatches.begin(); mati != fgpMatches.end(); mati++) {
//FIX: we will assume that the first atom in fgrp is always the connection
// atom
amat = mati->front();
aid = amat.second; //FIX: is this correct - the second entry in the pair is the atom ID from mol
// grab the list of atom Ids from mol that match the functional group
INT_VECT bondIds, maids;
for (mi = mati->begin(); mi != mati->end(); mi++) {
maids.push_back(mi->second);
}
// create a list of bond IDs from these atom ID
// these are the bond in mol that are part of portion that matches the
// functional group
bondIds = Subgraphs::bondListFromAtomList(mol, maids);
// now check if all these bonds have been covered as part of larger
// functional group that was dealt with earlier
// FIX: obviously we assume here that the function groups in params
// come in decreasing order of size.
bool allDone = true;
for (bi = bondIds.begin(); bi != bondIds.end(); bi++) {
if (std::find(fgBonds.begin(), fgBonds.end(), (*bi)) == fgBonds.end()) {
allDone = false;
fgBonds.push_back(*bi);
}
}
if (!allDone) {
// this functional group mapping onto mol is not part of a larger func
// group mapping so record it
aidFgrps.push_back(std::pair<int, int>(aid, fid));
}
}
fid++;
}
return aidFgrps;
}
bool fragmentMol(const ROMol& mol,
std::vector<std::pair<ROMOL_SPTR, ROMOL_SPTR> >& res,
unsigned int minCuts,
unsigned int maxCuts,
unsigned int maxCutBonds,
const std::string& pattern) {
#ifdef _DEBUG
for (size_t i = 0; i < mol.getNumAtoms(); i++) {
std::string symbol = mol.getAtomWithIdx(i)->getSymbol();
int label = 0;
mol.getAtomWithIdx(i)->getPropIfPresent(common_properties::molAtomMapNumber,
label);
char a1[32];
if (0 == label)
sprintf(a1, "\'%s\'", symbol.c_str(), label);
else
sprintf(a1, "\'%s:%u\'", symbol.c_str(), label);
std::cout << "Atom " << i << ": " << a1; //<<" Bonds:";
std::cout << "\n";
}
#endif
res.clear();
std::auto_ptr<const ROMol> smarts((const ROMol*)SmartsToMol(pattern));
std::vector<MatchVectType>
matching_atoms; // one bond per match ! with default pattern
unsigned int total = SubstructMatch(mol, *smarts, matching_atoms);
#ifdef _DEBUG
std::cout << "total substructs =" << total
<< "\nmatching bonds (atom1, atom2):\n";
#endif
if (0 == total) // Not found. Return empty set of molecules
return false;
#ifdef _DEBUG
for (size_t i = 0; i < matching_atoms.size(); i++) {
std::string symbol =
mol.getAtomWithIdx(matching_atoms[i][0].second)->getSymbol();
int label = 0;
mol.getAtomWithIdx(matching_atoms[i][0].second)
->getPropIfPresent(common_properties::molAtomMapNumber, label);
char a1[32];
if (0 == label)
sprintf(a1, "\'%s\'", symbol.c_str(), label);
else
sprintf(a1, "\'%s:%u\'", symbol.c_str(), label);
symbol = mol.getAtomWithIdx(matching_atoms[i][1].second)->getSymbol();
label = 0;
mol.getAtomWithIdx(matching_atoms[i][1].second)
->getPropIfPresent(common_properties::molAtomMapNumber, label);
char a2[32];
if (0 == label)
sprintf(a2, "\'%s\'", symbol.c_str(), label);
else
sprintf(a2, "\'%s:%u\'", symbol.c_str(), label);
std::cout << i << ": (" << matching_atoms[i][0].second << a1 << ","
<< matching_atoms[i][1].second << a2 << ") \n";
}
#endif
std::vector<BondVector_t> matching_bonds; // List of matched query's bonds
convertMatchingToBondVect(matching_bonds, matching_atoms, mol);
if (matching_bonds.size() > maxCutBonds) return false;
#ifdef _DEBUG
std::cout << "total matching_bonds = " << matching_bonds.size() << "\n";
#endif
// loop to generate every cut in the molecule
BondVector_t bonds_selected;
processCuts(0, minCuts, maxCuts, bonds_selected, matching_bonds, mol, res);
return true;
}
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;
}
// caller owns the result, it must be deleted
ExplicitBitVect *PatternFingerprintMol(const ROMol &mol, unsigned int fpSize,
std::vector<unsigned int> *atomCounts,
ExplicitBitVect *setOnlyBits) {
PRECONDITION(fpSize != 0, "fpSize==0");
PRECONDITION(!atomCounts || atomCounts->size() >= mol.getNumAtoms(),
"bad atomCounts size");
PRECONDITION(!setOnlyBits || setOnlyBits->getNumBits() == fpSize,
"bad setOnlyBits size");
std::vector<const ROMol *> patts;
patts.reserve(10);
unsigned int idx = 0;
while (1) {
std::string pq = pqs[idx];
if (pq == "") break;
++idx;
const ROMol *matcher = pattern_flyweight(pq).get().getMatcher();
CHECK_INVARIANT(matcher, "bad smarts");
patts.push_back(matcher);
}
if (!mol.getRingInfo()->isInitialized()) {
MolOps::fastFindRings(mol);
}
boost::dynamic_bitset<> isQueryAtom(mol.getNumAtoms()),
isQueryBond(mol.getNumBonds());
ROMol::VERTEX_ITER firstA, lastA;
boost::tie(firstA, lastA) = mol.getVertices();
while (firstA != lastA) {
const Atom *at = mol[*firstA].get();
if (isComplexQuery(at)) {
isQueryAtom.set(at->getIdx());
// std::cerr<<" complex atom: "<<at->getIdx()<<std::endl;
}
++firstA;
}
ROMol::EDGE_ITER firstB, lastB;
boost::tie(firstB, lastB) = mol.getEdges();
while (firstB != lastB) {
const Bond *bond = mol[*firstB].get();
// if( isComplexQuery(bond) ){
if (isPatternComplexQuery(bond)) {
isQueryBond.set(bond->getIdx());
// std::cerr<<" complex bond: "<<bond->getIdx()<<std::endl;
}
++firstB;
}
ExplicitBitVect *res = new ExplicitBitVect(fpSize);
unsigned int pIdx = 0;
BOOST_FOREACH (const ROMol *patt, patts) {
++pIdx;
std::vector<MatchVectType> matches;
// uniquify matches?
// time for 10K molecules w/ uniquify: 5.24s
// time for 10K molecules w/o uniquify: 4.87s
SubstructMatch(mol, *patt, matches, false);
boost::uint32_t mIdx = pIdx + patt->getNumAtoms() + patt->getNumBonds();
BOOST_FOREACH (MatchVectType &mv, matches) {
#ifdef VERBOSE_FINGERPRINTING
std::cerr << "\nPatt: " << pIdx << " | ";
#endif
// collect bits counting the number of occurances of the pattern:
gboost::hash_combine(mIdx, 0xBEEF);
res->setBit(mIdx % fpSize);
#ifdef VERBOSE_FINGERPRINTING
std::cerr << "count: " << mIdx % fpSize << " | ";
#endif
bool isQuery = false;
boost::uint32_t bitId = pIdx;
std::vector<unsigned int> amap(mv.size(), 0);
BOOST_FOREACH (MatchVectType::value_type &p, mv) {
#ifdef VERBOSE_FINGERPRINTING
std::cerr << p.second << " ";
#endif
if (isQueryAtom[p.second]) {
isQuery = true;
#ifdef VERBOSE_FINGERPRINTING
std::cerr << "atom query.";
#endif
break;
}
gboost::hash_combine(bitId,
mol.getAtomWithIdx(p.second)->getAtomicNum());
amap[p.first] = p.second;
}
