void ReadFileThread::run()
{
// Check that the file can be read from disk
if (!MoleculeFile::canOpen(m_moleculeFile->m_fileName, QFile::ReadOnly | QFile::Text)) {
// Cannot read the file
m_moleculeFile->m_error.append(QObject::tr("File %1 cannot be opened for reading.")
.arg(m_moleculeFile->m_fileName));
return;
}
// Construct the OpenBabel objects, set the file type
OpenBabel::OBConversion conv;
OpenBabel::OBFormat *inFormat;
if (!m_moleculeFile->m_fileType.isEmpty() &&
!conv.SetInFormat(m_moleculeFile->m_fileType.toAscii().data())) {
// Input format not supported
m_moleculeFile->m_error.append(
QObject::tr("File type '%1' is not supported for reading.")
.arg(m_moleculeFile->m_fileType));
return;
}
else {
inFormat = conv.FormatFromExt(m_moleculeFile->m_fileName.toAscii().data());
if (!inFormat || !conv.SetInFormat(inFormat)) {
// Input format not supported
m_moleculeFile->m_error
.append(QObject::tr("File type for file '%1' is not supported for reading.")
.arg(m_moleculeFile->m_fileName));
return;
}
}
// set any options
if (!m_moleculeFile->m_fileOptions.isEmpty()) {
foreach(const QString &option, m_moleculeFile
->m_fileOptions.split('\n', QString::SkipEmptyParts)) {
conv.AddOption(option.toAscii().data(), OBConversion::INOPTIONS);
}
}
//
// Read the input data files
//
void readInputFiles(int argc, char** argv,
std::vector<Linker>& linkers, std::vector<Rigid>& rigids)
{
// Input parser functionality for Open babel
OpenBabel::OBConversion obConversion;
obConversion.SetInFormat("sdf");
//
// For each file
//
for (int f = 1; f < argc; f++)
{
//
// For each molecule in this file.
//
OpenBabel::OBMol* mol = new OpenBabel::OBMol();
bool notatend = obConversion.ReadFile(mol, argv[f]);
while (notatend)
{
//
// Create this particular molecule type.
//
// Is there a better way to handle distinguising linkers vs. rigids
// other than the name of the file (prefix)?
switch(argv[f][0])
{
case 'l':
linkers.push_back(*(new Linker(*mol)));
break;
case 'r':
rigids.push_back(*(new Rigid(*mol)));
break;
default:
cerr << "Unexpected file prefix: " << argv[f][0] << " with file " << argv[f][0] << endl;
}
//
// Read the next molecule in from the file.
//
mol = new OpenBabel::OBMol();
notatend = obConversion.Read(mol);
}
}
}
Data::Geometry* ZMatrixCoordinates::parse(QString const& str)
{
OpenBabel::OBConversion conv;
conv.SetInFormat("gzmat");
// create dummy z-matrix input
std::string s("#\n\nzmat\n\n0 1\n");
s += str.toStdString();
OpenBabel::OBMol mol;
std::istringstream iss(s);
conv.Read(&mol, &iss);
Data::Geometry* geometry(new Data::Geometry);
for (::OpenBabel::OBMolAtomIter obAtom(&mol); obAtom; ++obAtom) {
qglviewer::Vec position(obAtom->x(), obAtom->y(), obAtom->z());
unsigned Z(obAtom->GetAtomicNum());
geometry->append(Z, position);
}
geometry->computeGasteigerCharges();
return geometry;
}
//*--------------------------------------------------------------------------*//
//* MAIN MAIN *//
//*--------------------------------------------------------------------------*//
int main(int argc, char* argv[])
{
// Initialise random number generator
srandom(time(NULL));
clock_t t0 = clock();
// Print header
printHeader();
// Read options
Options uo = parseCommandLine(argc,argv);
if(!uo.noHybrid)
{
if(uo.funcGroupVec[AROM] && uo.funcGroupVec[LIPO])
{
uo.funcGroupVec[HYBL] = true;
}
if(uo.funcGroupVec[HDON] && uo.funcGroupVec[HACC])
{
uo.funcGroupVec[HYBH] = true;
}
}
std::cerr << uo.print() << std::endl;
if (uo.version)
{
printHeader();
exit(0);
}
if (uo.help)
{
printUsage();
exit(0);
}
// Db file and pharmacophore out are mandatory elements
if (uo.dbInpFile.empty())
{
mainErr("Missing database file. This is a required option (-d).");
}
if (uo.pharmOutFile.empty() && uo.molOutFile.empty() && uo.scoreOutFile.empty())
{
mainErr("No output file defined. So there is actually no use to compute anything at all.");
}
if ((uo.pharmOutFile.empty() && uo.scoreOutFile.empty()) && !uo.molOutFile.empty())
{
mainErr("No file defined to write pharmacophore information.");
}
if (uo.refInpFile.empty() && uo.pharmOutFile.empty() && uo.molOutFile.empty() && !uo.scoreOutFile.empty())
{
mainErr("Only score file requested when no reference is given. Unable to generate this output.");
}
// Reference variables
Pharmacophore refPharm;
refPharm.clear();
std::string refId;
double refVolume(0.0);
int refSize(0);
int exclSize(0);
// Database variables
std::vector<Result*> resList;
Pharmacophore dbPharm;
std::string dbId;
double dbVolume(0.0);
int dbSize(0);
//----------------------------------------------------------------------------
//...(A).. Process the reference
//----------------------------------------------------------------------------
if (!uo.refInpFile.empty())
{
//-------------------------------------------------------
//...(1).. get reference pharmacophore
//-------------------------------------------------------
if (uo.refInpType == UNKNOWN)
{
std::string ext(getExt(uo.refInpFile));
if (ext == ".phar")
{
uo.refInpType = PHAR;
}
else
{
uo.refInpType = MOL;
}
}
if (uo.refInpType == MOL)
{
OpenBabel::OBMol m;
OpenBabel::OBConversion* reader = new OpenBabel::OBConversion();
reader->SetInFormat(reader->FormatFromExt(uo.refInpFile.c_str()));
if (!reader->Read(&m, uo.refInpStream))
{
mainErr("Unable to read reference molecule");
}
calcPharm(&m, &refPharm, uo);
refId = m.GetTitle();
delete reader;
reader = NULL;
}
else if (uo.refInpType == PHAR)
{
PharmacophoreReader* reader = new PharmacophoreReader();
refPharm = reader->read(uo.refInpStream, refId);
if (refPharm.empty())
{
mainErr("Error reading reference pharmacophore");
}
delete reader;
reader = NULL;
}
else
{
mainErr("Unknown format of reference molecule.");
}
//-------------------------------------------------------
//...(2).. process reference pharmacophore
//-------------------------------------------------------
if (uo.merge)
{
pharMerger.merge(refPharm);
}
refSize = refPharm.size();
for (unsigned int i(0); i < refSize; ++i)
{
if (refPharm[i].func == EXCL)
{
// extract overlap with exclusion spheres
for (unsigned int j(0); j < refPharm.size(); ++j)
{
if (refPharm[j].func != EXCL)
{
refVolume -= VolumeOverlap(refPharm[i], refPharm[j], !uo.noNormal);
}
}
exclSize++;
}
else
{
// add point self-overlap
refVolume += VolumeOverlap(refPharm[i], refPharm[i], !uo.noNormal);
}
}
if(!uo.isQuiet)
{
std::cerr << "Reference pharmacophore " << refId << std::endl;
std::cerr << " number of points: " << refSize - exclSize << std::endl;
std::cerr << " number of exclusion spheres: " << exclSize << std::endl;
std::cerr << " totalvolume: " << refVolume << std::endl;
}
}
//----------------------------------------------------------------------------
//...(B).. Process the database file
//----------------------------------------------------------------------------
// DB files
if (uo.dbInpType == UNKNOWN)
{
std::string ext(getExt(uo.dbInpFile));
if (ext==".phar")
{
uo.dbInpType = PHAR;
}
else
{
uo.dbInpType = MOL;
}
}
// local storage of the rotation matrix
SiMath::Matrix rotMat(3,3,0.0);
unsigned int molCount(0);
OpenBabel::OBConversion* molReader = NULL;
PharmacophoreReader* pharmReader = NULL;
if (uo.dbInpType == PHAR)
{
pharmReader = new PharmacophoreReader();
}
else if (uo.dbInpType == MOL)
{
molReader = new OpenBabel::OBConversion();
molReader->SetInFormat(molReader->FormatFromExt(uo.dbInpFile.c_str()));
molReader->SetInStream(uo.dbInpStream);
}
else
{
mainErr("Unknown format of db file.");
}
bool done(false);
OpenBabel::OBMol m;
while (!done)
{
dbPharm.clear();
m.Clear();
if (uo.dbInpType == MOL)
{
if (!molReader->Read(&m))
{
done = true;
break;
}
else
{
calcPharm(&m, &dbPharm, uo);
dbId = m.GetTitle();
}
}
else
{
if (uo.dbInpStream->eof())
{
done = true;
break;
}
else
{
dbPharm = pharmReader->read(uo.dbInpStream, dbId);
}
}
if (dbPharm.empty())
{
continue;
}
++molCount;
if (!uo.isQuiet )
{
if ((molCount % 10) == 0)
{
std::cerr << "." << std::flush;
}
if ((molCount % 500) == 0)
{
std::cerr << molCount << std::endl << std::flush;
}
}
if (uo.merge)
{
pharMerger.merge(dbPharm);
}
if (uo.refInpFile.empty())
{
if (!(uo.isQuiet))
{
printProgress(molCount);
}
if( !uo.pharmOutFile.empty())
{
uo.pharmOutWriter->write(dbPharm, uo.pharmOutStream, dbId);
}
continue;
}
//-------------------------------------------------------
//...(1).. Alignment
//-------------------------------------------------------
dbSize = dbPharm.size();
dbVolume = 0.0;
for (unsigned int i(0); i < dbSize; ++i)
{
if (dbPharm[i].func == EXCL)
{
continue;
}
dbVolume += VolumeOverlap(dbPharm[i], dbPharm[i], !uo.noNormal);
}
// Create a result structure
Result res;
res.refId = refId;
res.refVolume = refVolume;
res.dbId = dbId;
res.dbVolume = dbVolume;
res.overlapVolume = 0.0;
res.exclVolume = 0.0;
res.resMol = m;
res.resPharSize = 0;
if (uo.scoreOnly)
{
FunctionMapping funcMap(&refPharm, &dbPharm, uo.epsilon);
PharmacophoreMap fMap = funcMap.getNextMap();
double volBest(-9999.999);
// loop over all reference points
while (!fMap.empty())
{
double newVol(0.0);
double exclVol(0.0);
for (PharmacophoreMap::iterator itP = fMap.begin(); itP != fMap.end(); ++itP)
{
if ((itP->first)->func == EXCL)
{
exclVol += VolumeOverlap((itP->first), (itP->second), !uo.noNormal);
}
else if (((itP->first)->func == (itP->second)->func ) ||
(((itP->first)->func == HYBH ||
(itP->first)->func == HDON ||
(itP->first)->func == HACC)
&& ((itP->second)->func == HDON ||
(itP->second)->func == HACC ||
(itP->second)->func == HYBH))
|| (((itP->first)->func == HYBL ||
(itP->first)->func == AROM ||
(itP->first)->func == LIPO)
&& ((itP->second)->func == AROM ||
(itP->second)->func == LIPO ||
(itP->second)->func == HYBL)))
{
newVol += VolumeOverlap((itP->first),(itP->second), !uo.noNormal);
}
}
if ((newVol - exclVol) > volBest)
{
res.resPhar.clear();
res.resPharSize = 0;
for (PharmacophoreMap::iterator itP = fMap.begin(); itP != fMap.end(); ++itP)
{
// add point to resulting pharmacophore
PharmacophorePoint p(itP->second);
(res.resPhar).push_back(p);
++res.resPharSize;
}
res.overlapVolume = newVol;
res.exclVolume = exclVol;
volBest = newVol - exclVol;
}
// get the next map
fMap.clear();
fMap = funcMap.getNextMap();
}
}
else
{
FunctionMapping funcMap(&refPharm, &dbPharm, uo.epsilon);
PharmacophoreMap fMap = funcMap.getNextMap();
PharmacophoreMap bestMap;
// default solution
SolutionInfo best;
best.volume = -999.9;
// rotor is set to no rotation
best.rotor.resize(4);
best.rotor = 0.0;
best.rotor[0] = 1.0;
double bestScore = -1000;
int mapSize(fMap.size());
int maxSize = mapSize - 3;
while (!fMap.empty())
{
int msize = fMap.size();
// add the exclusion spheres to the alignment procedure
if (uo.withExclusion)
{
for (unsigned int i(0); i < refSize ; ++i)
{
if (refPharm[i].func != EXCL)
{
continue;
}
for (unsigned int j(0); j < dbSize; ++j)
{
if (dbPharm[j].func == EXCL)
{
continue;
}
fMap.insert(std::make_pair(&(refPharm[i]), &(dbPharm[j])));
}
}
}
// Only align if the expected score has any chance of being larger
// than best score so far
if ((msize > maxSize)
&& (((double) msize / (refSize - exclSize + dbSize - msize)) > bestScore))
{
Alignment align(fMap);
SolutionInfo r = align.align(!uo.noNormal);
if (best.volume < r.volume)
{
best = r;
bestScore = best.volume / (refVolume + dbVolume - best.volume);
bestMap = fMap;
mapSize = msize;
}
}
else
{
// Level of mapping site to low
break;
}
if (bestScore > 0.98)
{
break;
}
// Get the next map
fMap.clear();
fMap = funcMap.getNextMap();
}
// Transform the complete pharmacophore and the molecule towards the
// best alignment
rotMat = quat2Rotation(best.rotor);
positionPharmacophore(dbPharm, rotMat, best);
positionMolecule(&res.resMol, rotMat, best);
// Update result
res.info = best;
// Compute overlap volume between exlusion spheres and pharmacophore
// points
for (int i(0); i < refSize; ++i)
{
if (refPharm[i].func != EXCL)
{
continue;
}
for (int j(0); j < dbSize; ++j)
{
res.exclVolume += VolumeOverlap(refPharm[i], dbPharm[j], !uo.noNormal);
}
}
// make copy of the best map and compute the volume overlap
for (PharmacophoreMap::iterator itP = bestMap.begin(); itP != bestMap.end(); ++itP)
{
if(((itP->first)->func == EXCL) || ((itP->second)->func == EXCL))
{
continue;
}
// compute overlap
res.overlapVolume += VolumeOverlap(itP->first, itP->second, !uo.noNormal);
// add point to resulting pharmacophore
PharmacophorePoint p(itP->second);
(res.resPhar).push_back(p);
++res.resPharSize;
}
}
// update scores
res.info.volume = res.overlapVolume - res.exclVolume;
if (res.info.volume > 0.0)
{
res.tanimoto = res.info.volume / (res.refVolume + res.dbVolume - res.info.volume);
res.tversky_ref = res.info.volume / res.refVolume;
res.tversky_db = res.info.volume / res.dbVolume;
}
switch (uo.rankby)
{
case TANIMOTO:
res.rankbyScore = res.tanimoto;
break;
case TVERSKY_REF:
res.rankbyScore = res.tversky_ref;
break;
case TVERSKY_DB:
res.rankbyScore = res.tversky_db;
break;
}
//-------------------------------------------------------
//...(5).. Generate output
//-------------------------------------------------------
if (uo.cutOff != 0.0)
{
if (res.rankbyScore < uo.cutOff)
{
continue;
}
}
if (uo.best != 0)
{
addBest(res, uo, resList);
}
else
{
if (!uo.molOutFile.empty())
{
logOut(&res, uo);
}
if (!uo.pharmOutFile.empty())
{
logPharmacophores(&res, uo);
}
if (!uo.scoreOutFile.empty())
{
logScores(&res, uo);
}
}
}
if (molReader)
{
delete molReader;
molReader = NULL;
}
if (pharmReader)
{
delete pharmReader;
pharmReader = NULL;
}
//----------------------------------------------------------------------------
//...(C).. Process best list (if defined)
//----------------------------------------------------------------------------
if (uo.best != 0)
{
std::vector<Result*>::iterator itR;
for (itR = resList.begin(); itR != resList.end(); ++itR)
{
Result* res(*itR);
if (!uo.molOutFile.empty())
{
logOut(res, uo);
}
if (!uo.pharmOutFile.empty())
{
logPharmacophores(res, uo);
}
if (!uo.scoreOutFile.empty())
{
logScores(res, uo);
}
delete res;
}
}
// done processing database
if (!uo.isQuiet)
{
if (uo.refInpFile.empty())
{
std::cerr << std::endl;
std::cerr << "Processed " << molCount << " molecules";
double tt = (double)(clock() - t0 )/CLOCKS_PER_SEC;
std::cerr << " in " << tt << " seconds (";
std::cerr << molCount/tt << " molecules per second)." << std::endl;
}
else
{
std::cerr << std::endl;
std::cerr << "Processed " << molCount << " molecules" << std::endl;
double tt = (double)(clock() - t0 )/CLOCKS_PER_SEC;
std::cerr << molCount << " alignments in " << tt << " seconds (";
std::cerr << molCount/tt << " alignments per second)." << std::endl;
}
}
exit(0);
}
void readMoleculeFile(const char* fileName)
{
//
// Input parser conversion functionality for Open babel
//
OpenBabel::OBConversion obConversion;
obConversion.SetInFormat("SDF");
//
// Open the file, split the current molecule into Molecule Data (prefix)
// and Our Data (Suffix)
//
std::ifstream infile;
infile.open(fileName);
std::string name = "UNKNOWN";
std::string prefix = "";
std::string suffix = "";
while(splitMolecule(infile, name, prefix, suffix))
{
//
// If the name of molecule is not given, overwrite it with the name of the file.
//
if (name == "UNKNOWN")
{
name = "#### ";
name += fileName;
name += " ####";
}
if (g_debug_output) std::cerr << "Name: " << std::endl << name << std::endl;
if (g_debug_output) std::cerr << "Prefix: " << std::endl << prefix << std::endl;
if (g_debug_output) std::cerr << "Suffix: " << std::endl << suffix << std::endl;
// Create and parse using Open Babel
OpenBabel::OBMol* mol = new OpenBabel::OBMol();
bool notAtEnd = obConversion.ReadString(mol, prefix);
// Assign all needed data to the molecule (comment data)
Molecule* local = createLocalMolecule(mol, fileName[0] == 'l' ? LINKER : RIGID,
name, suffix);
// calculate the molecular weight, H donors and acceptors and the plogp
local->openBabelPredictLipinski();
// add to logfile
if (local->islipinskiPredicted())
{
std::ofstream logfile("synth_log_initial_fragments_logfile.txt",
std::ofstream::out | std::ofstream::app); // append
logfile << fileName << "\nMolWt = " << local->getMolWt() << "\n";
logfile << "HBD = " << local->getHBD() << "\n";
logfile << "HBA1 = " << local->getHBA1() << "\n";
logfile << "logP = " << local->getlogP() << "\n";
logfile << std::endl;
logfile.close();
}
else std::cerr << "Main: predictLipinski failed somehow!" << endl;
if (g_debug_output) std::cout << "Local: " << *local << "|" << std::endl;
// Add to the linker or rigid list as needed.
addMolecule(fileName[0], local);
}
}
int main(int argc,char **argv)
{
std::string ifile = "";
OpenBabel::OBSpectrophore::AccuracyOption accuracy = OpenBabel::OBSpectrophore::AngStepSize20;
OpenBabel::OBSpectrophore::StereoOption stereo = OpenBabel::OBSpectrophore::NoStereoSpecificProbes;
OpenBabel::OBSpectrophore::NormalizationOption normalization = OpenBabel::OBSpectrophore::NoNormalization;
double resolution = 3.0;
int c;
opterr = 0;
std::string msg;
while ((c = getopt(argc, argv, "ui:n:a:s:r:h")) != -1)
{
switch (c)
{
case 'u':
showImplementationDetails(argv[0]);
exit(1);
break;
case 'i':
if (!isValidValue('i', optarg))
{
msg = "Option -i is followed by an invalid argument: ";
msg += optarg;
showError(msg);
exit(1);
}
else
{
ifile = optarg;
}
break;
case 'n':
if (!isValidValue('n', optarg))
{
msg = "Option -n is followed by an invalid argument: ";
msg += optarg;
showError(msg);
exit(1);
}
else normalization = stringToNormalizationOption(optarg);
break;
case 'a':
if (!isValidValue('a', optarg))
{
msg = "Option -a is followed by an invalid argument: ";
msg += optarg;
showError(msg);
exit(1);
}
else accuracy = stringToAccuracyOption(optarg);
break;
case 's':
if (!isValidValue('s', optarg))
{
msg = "Option -s is followed by an invalid argument: ";
msg += optarg;
showError(msg);
exit(1);
}
else stereo = stringToStereoOption(optarg);
break;
case 'r':
if (!isValidValue('r', optarg))
{
msg = "Option -r is followed by an invalid argument: ";
msg += optarg;
showError(msg);
exit(1);
}
else
{
resolution = atof(optarg);
if (resolution <= 0)
{
msg = "Resolution -r should be larger than 0.0: ";
msg += optarg;
showError(msg);
exit(1);
}
}
break;
case 'h':
showHelp(argv[0]);
exit(0);
break;
case '?':
if ((optopt == 'i') ||
(optopt == 'n') ||
(optopt == 'a') ||
(optopt == 's') ||
(optopt == 'r'))
{
msg = "Option -";
msg += optopt;
msg += " requires an argument.";
showError(msg);
exit(1);
}
else
{
msg = "Unknown option -";
msg += optopt;
msg += ".";
showError(msg);
exit(1);
}
break;
default:
showError("Unknown option");
exit(1);
break;
}
}
// The input file (-i) is the only required option
if (ifile.empty())
{
msg = "Input file specification is required (option -i).";
showError(msg);
exit(1);
}
OpenBabel::OBConversion obconversion;
OpenBabel::OBFormat *format = obconversion.FormatFromExt(ifile.c_str());
if (!format)
{
msg = "Could not find file format for ";
msg += ifile;
showError(msg);
exit(1);
}
obconversion.SetInFormat(format);
std::ifstream ifs;
ifs.open(ifile.c_str());
obconversion.SetInStream(&ifs);
// Start calculations
OpenBabel::OBMol mol;
OpenBabel::OBSpectrophore spec;
spec.SetAccuracy(accuracy);
spec.SetNormalization(normalization);
spec.SetStereo(stereo);
spec.SetResolution(resolution);
showParameters(spec, ifile);
unsigned int count(0);
while (obconversion.Read(&mol))
{
std::vector<double> result = spec.GetSpectrophore(&mol);
if (result.empty()) {
std::cerr << "Error calculating Spectrophore from molecule number ";
std::cerr << count;
std::cerr << " (counting starts at 0)!";
std::cerr << std::endl;
}
else
{
std::cout << mol.GetTitle() << "\t";
for (unsigned int i(0); i < result.size(); ++i)
{
std::cout << result[i] << "\t";
}
std::cout << std::endl;
}
mol.Clear();
++count;
}
return 0;
}