void OBAtomTyper::AssignTypes(OBMol &mol)
{
if (!_init)
Init();
obErrorLog.ThrowError(__FUNCTION__,
"Ran OpenBabel::AssignTypes", obAuditMsg);
mol.SetAtomTypesPerceived();
vector<vector<int> >::iterator j;
vector<pair<OBSmartsPattern*,string> >::iterator i;
for (i = _vexttyp.begin();i != _vexttyp.end();++i)
if (i->first->Match(mol))
{
_mlist = i->first->GetMapList();
for (j = _mlist.begin();j != _mlist.end();++j)
mol.GetAtom((*j)[0])->SetType(i->second);
}
// Special cases
vector<OBAtom*>::iterator a;
OBAtom* atom;
for (atom = mol.BeginAtom(a); atom; atom = mol.NextAtom(a)) {
// guanidinium. Fixes PR#1800964
if (strncasecmp(atom->GetType(),"C2", 2) == 0) {
int guanidineN = 0;
OBAtom *nbr;
vector<OBBond*>::iterator k;
for (nbr = atom->BeginNbrAtom(k);nbr;nbr = atom->NextNbrAtom(k)) {
if (strncasecmp(nbr->GetType(),"Npl", 3) == 0 ||
strncasecmp(nbr->GetType(),"N2", 2) == 0 ||
strncasecmp(nbr->GetType(),"Ng+", 3) == 0)
++guanidineN;
}
if (guanidineN == 3)
atom->SetType("C+");
} // end C2 carbon for guanidinium
} // end special cases
}
bool PDBFormat::WriteMolecule(OBBase* pOb, OBConversion* pConv)
{
OBMol* pmol = dynamic_cast<OBMol*>(pOb);
if(pmol==NULL)
return false;
//Define some references so we can use the old parameter names
ostream &ofs = *pConv->GetOutStream();
OBMol &mol = *pmol;
unsigned int i;
char buffer[BUFF_SIZE];
char type_name[10], padded_name[10];
char the_res[10];
char the_chain = ' ';
const char *element_name;
int res_num;
bool het=true;
int model_num = 0;
if (!pConv->IsLast() || pConv->GetOutputIndex() > 1)
{ // More than one molecule record
model_num = pConv->GetOutputIndex(); // MODEL 1-based index
snprintf(buffer, BUFF_SIZE, "MODEL %8d", model_num);
ofs << buffer << endl;
}
// write back all fields (REMARKS, HELIX, SHEET, SITE, ...)
bool compndWritten = false;
bool authorWritten = false;
std::vector<OBGenericData*> pairData = mol.GetAllData(OBGenericDataType::PairData);
for (std::vector<OBGenericData*>::iterator data = pairData.begin(); data != pairData.end(); ++data) {
OBPairData *pd = static_cast<OBPairData*>(*data);
string attr = pd->GetAttribute();
// filter to make sure we are writing pdb fields only
if (attr != "HEADER" && attr != "OBSLTE" && attr != "TITLE" && attr != "SPLIT" &&
attr != "CAVEAT" && attr != "COMPND" && attr != "SOURCE" && attr != "KEYWDS" &&
attr != "EXPDTA" && attr != "NUMMDL" && attr != "MDLTYP" && attr != "AUTHOR" &&
attr != "REVDAT" && attr != "SPRSDE" && attr != "JRNL" && attr != "REMARK" &&
attr != "DBREF" && attr != "DBREF1" && attr != "DBREF2" && attr != "SEQADV" &&
attr != "SEQRES" && attr != "MODRES" && attr != "HET" && attr != "HETNAM" &&
attr != "HETSYN" && attr != "FORMUL" && attr != "HELIX" && attr != "SHEET" &&
attr != "SSBOND" && attr != "LINK" && attr != "CISPEP" && attr != "SITE" &&
attr != "ORIGX1" && attr != "ORIGX2" && attr != "ORIGX3" && attr != "SCALE1" &&
attr != "SCALE2" && attr != "SCALE3" && attr != "MATRIX1" && attr != "MATRIX2" &&
attr != "MATRIX3" && attr != "MODEL")
continue;
if (attr == "COMPND")
compndWritten = true;
if (attr == "AUTHOR")
authorWritten = true;
// compute spacing needed. HELIX, SITE, HET, ... are trimmed when reading
int nSpacing = 6 - attr.size();
for (int i = 0; i < nSpacing; ++i)
attr += " ";
std::string lines = pd->GetValue();
string::size_type last = 0;
string::size_type pos = lines.find('\n');
while (last != string::npos) {
string line = lines.substr(last, pos - last);
if (pos == string::npos)
last = string::npos;
else
last = pos + 1;
pos = lines.find('\n', last);
ofs << attr << line << endl;
}
}
if (!compndWritten) {
if (strlen(mol.GetTitle()) > 0)
snprintf(buffer, BUFF_SIZE, "COMPND %s ",mol.GetTitle());
else
snprintf(buffer, BUFF_SIZE, "COMPND UNNAMED");
ofs << buffer << endl;
}
if (!authorWritten) {
snprintf(buffer, BUFF_SIZE, "AUTHOR GENERATED BY OPEN BABEL %s",BABEL_VERSION);
ofs << buffer << endl;
}
// Write CRYST1 record, containing unit cell parameters, space group
// and Z value (supposed to be 1)
if (pmol->HasData(OBGenericDataType::UnitCell))
{
OBUnitCell *pUC = (OBUnitCell*)pmol->GetData(OBGenericDataType::UnitCell);
if(pUC->GetSpaceGroup()){
string tmpHM=pUC->GetSpaceGroup()->GetHMName();
// Do we have an extended HM symbol, with origin choice as ":1" or ":2" ? If so, remove it.
size_t n=tmpHM.find(":");
if(n!=string::npos) tmpHM=tmpHM.substr(0,n);
snprintf(buffer, BUFF_SIZE,
"CRYST1%9.3f%9.3f%9.3f%7.2f%7.2f%7.2f %-11s 1",
pUC->GetA(), pUC->GetB(), pUC->GetC(),
pUC->GetAlpha(), pUC->GetBeta(), pUC->GetGamma(),
tmpHM.c_str());
}
else
snprintf(buffer, BUFF_SIZE,
"CRYST1%9.3f%9.3f%9.3f%7.2f%7.2f%7.2f %-11s 1",
pUC->GetA(), pUC->GetB(), pUC->GetC(),
pUC->GetAlpha(), pUC->GetBeta(), pUC->GetGamma(),
"P1");
ofs << buffer << endl;
}
// before we write any records, we should check to see if any coord < -1000
// which will cause errors in the formatting
double minX, minY, minZ;
minX = minY = minZ = -999.0f;
FOR_ATOMS_OF_MOL(a, mol)
{
if (a->GetX() < minX)
minX = a->GetX();
if (a->GetY() < minY)
minY = a->GetY();
if (a->GetZ() < minZ)
minZ = a->GetZ();
}
vector3 transV = VZero;
if (minX < -999.0)
transV.SetX(-1.0*minX - 900.0);
if (minY < -999.0)
transV.SetY(-1.0*minY - 900.0);
if (minZ < -999.0)
transV.SetZ(-1.0*minZ - 900.0);
// if minX, minY, or minZ was never changed, shift will be 0.0f
// otherwise, move enough so that smallest coord is > -999.0f
mol.Translate(transV);
OBAtom *atom;
OBResidue *res;
for (i = 1; i <= mol.NumAtoms(); i++)
{
atom = mol.GetAtom(i);
strncpy(type_name, etab.GetSymbol(atom->GetAtomicNum()), sizeof(type_name));
type_name[sizeof(type_name) - 1] = '\0';
//two char. elements are on position 13 and 14 one char. start at 14
if (strlen(type_name) > 1)
type_name[1] = toupper(type_name[1]);
else
{
char tmp[10];
strncpy(tmp, type_name, 9); // make sure to null-terminate tmp
snprintf(type_name, sizeof(type_name), " %-3s", tmp);
}
if ( (res = atom->GetResidue()) != 0 )
{
het = res->IsHetAtom(atom);
snprintf(the_res,4,"%s",(char*)res->GetName().c_str());
the_res[4] = '\0';
snprintf(type_name,5,"%s",(char*)res->GetAtomID(atom).c_str());
the_chain = res->GetChain();
//two char. elements are on position 13 and 14 one char. start at 14
if (strlen(etab.GetSymbol(atom->GetAtomicNum())) == 1)
{
if (strlen(type_name) < 4)
{
char tmp[10];
strncpy(tmp, type_name, 9); // make sure to null-terminate tmp
snprintf(padded_name, sizeof(padded_name), " %-3s", tmp);
strncpy(type_name,padded_name,4);
type_name[4] = '\0';
}
else
{
/*
type_name[4] = type_name[3];
type_name[3] = type_name[2];
type_name[2] = type_name[1];
type_name[1] = type_name[0];
type_name[0] = type_name[4];
*/
type_name[4] = '\0';
}
}
res_num = res->GetNum();
}
else
{
strcpy(the_res,"UNK");
the_res[3] = '\0';
snprintf(padded_name,sizeof(padded_name), "%s",type_name);
strncpy(type_name,padded_name,4);
type_name[4] = '\0';
res_num = 1;
}
element_name = etab.GetSymbol(atom->GetAtomicNum());
int charge = atom->GetFormalCharge();
char scharge[3] = { ' ', ' ', '\0' };
if(0 != charge)
{
snprintf(scharge, 3, "%+d", charge);
char tmp = scharge[1];
scharge[1] = scharge[0];
scharge[0] = tmp;
}
snprintf(buffer, BUFF_SIZE, "%s%5d %-4s %-3s %c%4d %8.3f%8.3f%8.3f 1.00 0.00 %2s%2s\n",
het?"HETATM":"ATOM ",
i,
type_name,
the_res,
the_chain,
res_num,
atom->GetX(),
atom->GetY(),
atom->GetZ(),
element_name,
scharge);
ofs << buffer;
}
OBAtom *nbr;
vector<OBBond*>::iterator k;
for (i = 1; i <= mol.NumAtoms(); i ++)
{
atom = mol.GetAtom(i);
if (atom->GetValence() == 0)
continue; // no need to write a CONECT record -- no bonds
snprintf(buffer, BUFF_SIZE, "CONECT%5d", i);
ofs << buffer;
// Write out up to 4 real bonds per line PR#1711154
int currentValence = 0;
for (nbr = atom->BeginNbrAtom(k);nbr;nbr = atom->NextNbrAtom(k))
{
snprintf(buffer, BUFF_SIZE, "%5d", nbr->GetIdx());
ofs << buffer;
if (++currentValence % 4 == 0) {
// Add the trailing space to finish this record
ofs << " \n";
// write the start of a new CONECT record
snprintf(buffer, BUFF_SIZE, "CONECT%5d", i);
ofs << buffer;
}
}
// Add trailing spaces
int remainingValence = atom->GetValence() % 4;
for (int count = 0; count < (4 - remainingValence); count++) {
snprintf(buffer, BUFF_SIZE, " ");
ofs << buffer;
}
ofs << " \n";
}
snprintf(buffer, BUFF_SIZE, "MASTER 0 0 0 0 0 0 0 0 ");
ofs << buffer;
snprintf(buffer, BUFF_SIZE, "%4d 0 %4d 0\n",mol.NumAtoms(),mol.NumAtoms());
ofs << buffer;
ofs << "END\n";
if (model_num) {
ofs << "ENDMDL" << endl;
}
return(true);
}
bool BGFFormat::WriteMolecule(OBBase* pOb, OBConversion* pConv)
{
OBMol* pmol = dynamic_cast<OBMol*>(pOb);
if(pmol==NULL)
return false;
//Define some references so we can use the old parameter names
ostream &ofs = *pConv->GetOutStream();
OBMol &mol = *pmol;
vector<OBAtom*>::iterator i;
int max_val;
OBAtom *atom;
char buffer[BUFF_SIZE];
char elmnt_typ[8], dreid_typ[8], atm_sym[16], max_val_str[8];
mol.Kekulize();
ofs << "BIOGRF 200\n";
snprintf(buffer, BUFF_SIZE, "DESCRP %s\n",mol.GetTitle());
ofs << buffer;
snprintf(buffer, BUFF_SIZE, "REMARK BGF file created by Open Babel %s\n",BABEL_VERSION);
ofs << "FORCEFIELD DREIDING \n";
// write unit cell if available
if (mol.HasData(OBGenericDataType::UnitCell))
{
OBUnitCell *uc = (OBUnitCell*)mol.GetData(OBGenericDataType::UnitCell);
// e.g. CRYSTX 49.30287 49.23010 25.45631 90.00008 89.99995 57.10041
snprintf(buffer, BUFF_SIZE,
"CRYSTX%12.5f%12.5f%12.5f%12.5f%12.5f%12.5f",
uc->GetA(), uc->GetB(), uc->GetC(),
uc->GetAlpha() , uc->GetBeta(), uc->GetGamma());
ofs << buffer << "\n";
}
ofs << "FORMAT ATOM (a6,1x,i5,1x,a5,1x,a3,1x,a1,1x,a5,3f10.5,1x,a5,i3,i2,1x,f8.5)\n";
ttab.SetFromType("INT");
for (atom = mol.BeginAtom(i);atom;atom = mol.NextAtom(i))
{
strncpy(elmnt_typ,etab.GetSymbol(atom->GetAtomicNum()), 7); // make sure to null-terminate
elmnt_typ[sizeof(elmnt_typ) - 1] = '0';
ToUpper(elmnt_typ);
ttab.SetToType("DRE");
ttab.Translate(dreid_typ,atom->GetType());
ttab.SetToType("HAD");
ttab.Translate(max_val_str,atom->GetType());
max_val = atoi(max_val_str);
if (max_val == 0)
max_val = 1;
snprintf(atm_sym,16,"%s%d",elmnt_typ,atom->GetIdx());
snprintf(buffer,BUFF_SIZE,"%6s %5d %-5s %3s %1s %5s%10.5f%10.5f%10.5f %-5s%3d%2d %8.5f\n",
"HETATM",
atom->GetIdx(),
atm_sym,
"RES",
"A",
"444",
atom->GetX(),
atom->GetY(),
atom->GetZ(),
dreid_typ,
max_val,
0,
atom->GetPartialCharge());
ofs << buffer;
}
ofs<< "FORMAT CONECT (a6,12i6)\n\n";
OBAtom *nbr;
vector<OBBond*>::iterator j;
for (atom = mol.BeginAtom(i);atom;atom = mol.NextAtom(i))
if (atom->GetValence())
{
snprintf(buffer,BUFF_SIZE,"CONECT%6d",atom->GetIdx());
ofs << buffer;
for (nbr = atom->BeginNbrAtom(j);nbr;nbr = atom->NextNbrAtom(j))
{
snprintf(buffer,BUFF_SIZE,"%6d",nbr->GetIdx());
ofs << buffer;
}
ofs << endl;
snprintf(buffer,BUFF_SIZE,"ORDER %6d",atom->GetIdx());
ofs << buffer;
for (nbr = atom->BeginNbrAtom(j);nbr;nbr = atom->NextNbrAtom(j))
{
snprintf(buffer,BUFF_SIZE,"%6d",(*j)->GetBO());
ofs << buffer;
}
ofs << endl;
}
ofs << "END" << endl;
return(true);
}
bool PDBFormat::WriteMolecule(OBBase* pOb, OBConversion* pConv)
{
OBMol* pmol = dynamic_cast<OBMol*>(pOb);
if(pmol==NULL)
return false;
//Define some references so we can use the old parameter names
ostream &ofs = *pConv->GetOutStream();
OBMol &mol = *pmol;
unsigned int i;
char buffer[BUFF_SIZE];
char type_name[10], padded_name[10];
char the_res[10];
char the_chain = ' ';
const char *element_name;
int res_num;
bool het=true;
int model_num = 0;
if (!pConv->IsLast() || pConv->GetOutputIndex() > 1)
{ // More than one molecule record
model_num = pConv->GetOutputIndex(); // MODEL 1-based index
snprintf(buffer, BUFF_SIZE, "MODEL %8d", model_num);
ofs << buffer << endl;
}
if (strlen(mol.GetTitle()) > 0)
snprintf(buffer, BUFF_SIZE, "COMPND %s ",mol.GetTitle());
else
snprintf(buffer, BUFF_SIZE, "COMPND UNNAMED");
ofs << buffer << endl;
snprintf(buffer, BUFF_SIZE, "AUTHOR GENERATED BY OPEN BABEL %s",BABEL_VERSION);
ofs << buffer << endl;
// Write CRYST1 record, containing unit cell parameters, space group
// and Z value (supposed to be 1)
if (pmol->HasData(OBGenericDataType::UnitCell))
{
OBUnitCell *pUC = (OBUnitCell*)pmol->GetData(OBGenericDataType::UnitCell);
snprintf(buffer, BUFF_SIZE,
"CRYST1%9.3f%9.3f%9.3f%7.2f%7.2f%7.2f %-11s 1",
pUC->GetA(), pUC->GetB(), pUC->GetC(),
pUC->GetAlpha(), pUC->GetBeta(), pUC->GetGamma(),
pUC->GetSpaceGroup() ?
pUC->GetSpaceGroup()->GetHMName().c_str() : "P1");
ofs << buffer << endl;
}
// before we write any records, we should check to see if any coord < -1000
// which will cause errors in the formatting
double minX, minY, minZ;
minX = minY = minZ = -999.0f;
FOR_ATOMS_OF_MOL(a, mol)
{
if (a->GetX() < minX)
minX = a->GetX();
if (a->GetY() < minY)
minY = a->GetY();
if (a->GetZ() < minZ)
minZ = a->GetZ();
}
vector3 transV = VZero;
if (minX < -999.0)
transV.SetX(-1.0*minX - 900.0);
if (minY < -999.0)
transV.SetY(-1.0*minY - 900.0);
if (minZ < -999.0)
transV.SetZ(-1.0*minZ - 900.0);
// if minX, minY, or minZ was never changed, shift will be 0.0f
// otherwise, move enough so that smallest coord is > -999.0f
mol.Translate(transV);
OBAtom *atom;
OBResidue *res;
for (i = 1; i <= mol.NumAtoms(); i++)
{
atom = mol.GetAtom(i);
strncpy(type_name, etab.GetSymbol(atom->GetAtomicNum()), sizeof(type_name));
type_name[sizeof(type_name) - 1] = '\0';
//two char. elements are on position 13 and 14 one char. start at 14
if (strlen(type_name) > 1)
type_name[1] = toupper(type_name[1]);
else
{
char tmp[10];
strncpy(tmp, type_name, 10);
snprintf(type_name, sizeof(type_name), " %-3s", tmp);
}
if ( (res = atom->GetResidue()) != 0 )
{
het = res->IsHetAtom(atom);
snprintf(the_res,4,"%s",(char*)res->GetName().c_str());
snprintf(type_name,5,"%s",(char*)res->GetAtomID(atom).c_str());
the_chain = res->GetChain();
//two char. elements are on position 13 and 14 one char. start at 14
if (strlen(etab.GetSymbol(atom->GetAtomicNum())) == 1)
{
if (strlen(type_name) < 4)
{
char tmp[16];
strncpy(tmp, type_name, 16);
snprintf(padded_name, sizeof(padded_name), " %-3s", tmp);
strncpy(type_name,padded_name,4);
type_name[4] = '\0';
}
else
{
/*
type_name[4] = type_name[3];
type_name[3] = type_name[2];
type_name[2] = type_name[1];
type_name[1] = type_name[0];
type_name[0] = type_name[4];
*/
type_name[4] = '\0';
}
}
res_num = res->GetNum();
}
else
{
strcpy(the_res,"UNK");
snprintf(padded_name,sizeof(padded_name), "%s",type_name);
strncpy(type_name,padded_name,4);
type_name[4] = '\0';
res_num = 1;
}
element_name = etab.GetSymbol(atom->GetAtomicNum());
int charge = atom->GetFormalCharge();
char scharge[3] = { ' ', ' ', '\0' };
if(0 != charge)
{
snprintf(scharge, 3, "%+d", charge);
char tmp = scharge[1];
scharge[1] = scharge[0];
scharge[0] = tmp;
}
snprintf(buffer, BUFF_SIZE, "%s%5d %-4s %-3s %c%4d %8.3f%8.3f%8.3f 1.00 0.00 %2s%2s\n",
het?"HETATM":"ATOM ",
i,
type_name,
the_res,
the_chain,
res_num,
atom->GetX(),
atom->GetY(),
atom->GetZ(),
element_name,
scharge);
ofs << buffer;
}
OBAtom *nbr;
vector<OBBond*>::iterator k;
for (i = 1; i <= mol.NumAtoms(); i ++)
{
atom = mol.GetAtom(i);
if (atom->GetValence() == 0)
continue; // no need to write a CONECT record -- no bonds
snprintf(buffer, BUFF_SIZE, "CONECT%5d", i);
ofs << buffer;
// Write out up to 4 real bonds per line PR#1711154
unsigned int currentValence = 0;
for (nbr = atom->BeginNbrAtom(k);nbr;nbr = atom->NextNbrAtom(k))
{
unsigned int order = mol.GetBond(atom, nbr)->GetBondOrder();
unsigned int it_order = 0;
for( it_order = 0; it_order < order; it_order++ )
{
if (0 != currentValence && 0 == currentValence % 4)
{
// Add the trailing space to finish this record
ofs << " \n";
// write the start of a new CONECT record
snprintf(buffer, BUFF_SIZE, "CONECT%5d", i);
ofs << buffer;
}
currentValence++;
snprintf(buffer, BUFF_SIZE, "%5d", nbr->GetIdx());
ofs << buffer;
}
}
// Add trailing spaces
unsigned int remainingValence = currentValence % 4;
if( 0 < remainingValence )
{
for (int count = 0; count < (4 - remainingValence); count++)
{
snprintf(buffer, BUFF_SIZE, " ");
ofs << buffer;
}
}
ofs << " \n";
}
snprintf(buffer, BUFF_SIZE, "MASTER 0 0 0 0 0 0 0 0 ");
ofs << buffer;
snprintf(buffer, BUFF_SIZE, "%4d 0 %4d 0\n",mol.NumAtoms(),mol.NumAtoms());
ofs << buffer;
ofs << "END\n";
if (model_num) {
ofs << "ENDMDL" << endl;
}
return(true);
}