Beispiel #1
0
Subgraphs::DiscrimTuple FragCatalogEntry::getDiscrims() const {
  Subgraphs::DiscrimTuple res;
  if (this->hasProp(common_properties::Discrims)) {
    this->getProp(common_properties::Discrims, res);
  } else {
    PATH_TYPE path;
    for (unsigned int i = 0; i < dp_mol->getNumBonds(); ++i) path.push_back(i);

    // create invariant additions to reflect the functional groups attached to
    // the atoms
    std::vector<std::uint32_t> funcGpInvars;
    gboost::hash<INT_VECT> vectHasher;
    for (ROMol::AtomIterator atomIt = dp_mol->beginAtoms();
         atomIt != dp_mol->endAtoms(); ++atomIt) {
      unsigned int aid = (*atomIt)->getIdx();
      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.push_back(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;
}
Beispiel #2
0
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 = "";
}
Beispiel #3
0
std::vector<unsigned int> generateBondHashes(const ROMol &mol, boost::dynamic_bitset<>& atomsInPath,
    const std::vector<const Bond *>& bondCache,
    const std::vector<short>& isQueryBond,
    const PATH_TYPE &path, bool useBondOrder,
    std::vector<boost::uint32_t> *atomInvariants){

  PRECONDITION(!atomInvariants || atomInvariants->size() >= mol.getNumAtoms(),
               "bad atomInvariants size");

  std::vector<unsigned int> bondHashes;
  atomsInPath.reset();
  bool queryInPath=false;
  std::vector<unsigned int> atomDegrees(mol.getNumAtoms(),0);
  for(unsigned int i=0;i<path.size() && !queryInPath;++i){
    const Bond *bi = bondCache[path[i]];
    atomDegrees[bi->getBeginAtomIdx()]++;
    atomDegrees[bi->getEndAtomIdx()]++;
    atomsInPath.set(bi->getBeginAtomIdx());
    atomsInPath.set(bi->getEndAtomIdx());
    if(isQueryBond[path[i]]) queryInPath=true;
  }
  if(queryInPath){
    return bondHashes;
  }

  // -----------------
  // calculate the bond hashes:
  std::vector<unsigned int> bondNbrs(path.size(),0);
  bondHashes.reserve(path.size()+1);

  for(unsigned int i=0;i<path.size();++i){
    const Bond *bi = bondCache[path[i]];
#ifdef REPORT_FP_STATS
        if (std::find(atomsToUse.begin(), atomsToUse.end(),
                      bi->getBeginAtomIdx()) == atomsToUse.end()) {
          atomsToUse.push_back(bi->getBeginAtomIdx());
        }
        if (std::find(atomsToUse.begin(), atomsToUse.end(),
                      bi->getEndAtomIdx()) == atomsToUse.end()) {
          atomsToUse.push_back(bi->getEndAtomIdx());
        }
#endif
    for(unsigned int j=i+1;j<path.size();++j){
      const Bond *bj = bondCache[path[j]];
      if(bi->getBeginAtomIdx()==bj->getBeginAtomIdx() ||
          bi->getBeginAtomIdx()==bj->getEndAtomIdx() ||
          bi->getEndAtomIdx()==bj->getBeginAtomIdx() ||
          bi->getEndAtomIdx()==bj->getEndAtomIdx() ){
        ++bondNbrs[i];
        ++bondNbrs[j];
      }
    }
#ifdef VERBOSE_FINGERPRINTING
        std::cerr << "   bond(" << i << "):" << bondNbrs[i] << std::endl;
#endif
    // we have the count of neighbors for bond bi, compute its hash:
    unsigned int a1Hash = (*atomInvariants)[bi->getBeginAtomIdx()];
    unsigned int a2Hash = (*atomInvariants)[bi->getEndAtomIdx()];
    unsigned int deg1=atomDegrees[bi->getBeginAtomIdx()];
    unsigned int deg2=atomDegrees[bi->getEndAtomIdx()];
    if(a1Hash<a2Hash){
      std::swap(a1Hash,a2Hash);
      std::swap(deg1,deg2);
    }
    else if(a1Hash==a2Hash && deg1<deg2){
      std::swap(deg1,deg2);
    }
    unsigned int bondHash=1;
    if(useBondOrder){
      if(bi->getIsAromatic() || bi->getBondType()==Bond::AROMATIC){
        // makes sure aromatic bonds always hash as aromatic
        bondHash = Bond::AROMATIC;
      }
      else {
        bondHash = bi->getBondType();
      }
    }
    boost::uint32_t ourHash=bondNbrs[i];
    gboost::hash_combine(ourHash,bondHash);
    gboost::hash_combine(ourHash,a1Hash);
    gboost::hash_combine(ourHash,deg1);
    gboost::hash_combine(ourHash,a2Hash);
    gboost::hash_combine(ourHash,deg2);
    bondHashes.push_back(ourHash);
    //std::cerr<<"    "<<bi->getIdx()<<" "<<a1Hash<<"("<<deg1<<")"<<"-"<<a2Hash<<"("<<deg2<<")"<<" "<<bondHash<<" -> "<<ourHash<<std::endl;
  }
  return bondHashes;
}