Skip to content

Commit ac49793

Browse files
mhekkelmhekkel
committed
Fix loading embedded restraint data
1 parent 9927b50 commit ac49793

File tree

1 file changed

+118
-110
lines changed

1 file changed

+118
-110
lines changed

‎src/compound.cpp

Lines changed: 118 additions & 110 deletions
Original file line numberDiff line numberDiff line change
@@ -550,142 +550,150 @@ class local_compound_factory_impl : public compound_factory_impl
550550
for (const auto &[id, name, threeLetterCode, group] :
551551
file["comp_list"]["chem_comp"].rows<std::string, std::string, std::string, std::string>("id", "name", "three_letter_code", "group"))
552552
{
553-
// if (std::regex_match(group, peptideRx))
554-
// mKnownPeptides.insert(threeLetterCode);
555-
// else if (cif::iequals(group, "DNA") or cif::iequals(group, "RNA"))
556-
// mKnownBases.insert(threeLetterCode);
557-
558-
// Test if this compound is known in CCD
559553
auto &rdb = m_local_file["comp_" + id];
560554
if (rdb.empty())
561555
{
562556
std::cerr << "Missing data in restraint file for id " + id + '\n';
563557
continue;
564558
}
565559

566-
cif::datablock db(id);
560+
construct_compound(rdb, id, name, threeLetterCode, group);
561+
}
562+
}
567563

568-
float formula_weight = 0;
569-
int formal_charge = 0;
570-
std::map<std::string,size_t> formula_data;
564+
compound *create(const std::string &id) override;
571565

572-
for (size_t ord = 1; const auto &[atom_id, type_symbol, type, charge, x, y, z] :
573-
rdb["chem_comp_atom"].rows<std::string, std::string, std::string, int, float, float, float>(
574-
"atom_id", "type_symbol", "type", "charge", "x", "y", "z"))
575-
{
576-
auto atom = cif::atom_type_traits(type_symbol);
577-
formula_weight += atom.weight();
578-
579-
formula_data[type_symbol] += 1;
580-
581-
db["chem_comp_atom"].emplace({
582-
{ "comp_id", id },
583-
{ "atom_id", atom_id },
584-
{ "type_symbol", type_symbol },
585-
{ "charge", charge },
586-
{ "model_Cartn_x", x, 3 },
587-
{ "model_Cartn_y", y, 3 },
588-
{ "model_Cartn_z", z, 3 },
589-
{ "pdbx_ordinal", ord++ }
590-
});
591-
592-
formal_charge += charge;
593-
}
566+
private:
594567

595-
for (size_t ord = 1; const auto &[atom_id_1, atom_id_2, type, aromatic] :
596-
rdb["chem_comp_bond"].rows<std::string, std::string, std::string, bool>("atom_id_1", "atom_id_2", "type", "aromatic"))
597-
{
598-
std::string value_order("SING");
599-
600-
if (cif::iequals(type, "single") or cif::iequals(type, "sing"))
601-
value_order = "SING";
602-
else if (cif::iequals(type, "double") or cif::iequals(type, "doub"))
603-
value_order = "DOUB";
604-
else if (cif::iequals(type, "triple") or cif::iequals(type, "trip"))
605-
value_order = "TRIP";
606-
607-
db["chem_comp_bond"].emplace({
608-
{ "comp_id", id },
609-
{ "atom_id_1", atom_id_1 },
610-
{ "atom_id_2", atom_id_2 },
611-
{ "value_order", value_order },
612-
{ "pdbx_aromatic_flag", aromatic },
613-
// TODO: fetch stereo_config info from chem_comp_chir
614-
{ "pdbx_ordinal", ord++ }
615-
});
616-
}
568+
compound *construct_compound(const datablock &db, const std::string &id, const std::string &name, const std::string &three_letter_code, const std::string &group);
617569

618-
db.emplace_back(rdb["pdbx_chem_comp_descriptor"]);
570+
cif::file m_local_file;
571+
};
619572

620-
std::string formula;
621-
for (bool first = true; const auto &[symbol, count]: formula_data)
573+
compound *local_compound_factory_impl::create(const std::string &id)
574+
{
575+
compound *result = nullptr;
576+
577+
for (auto &db : m_local_file)
578+
{
579+
if (db.name() == "comp_" + id)
580+
{
581+
auto chem_comp = db.get("chem_comp");
582+
if (not chem_comp)
583+
break;
584+
585+
try
622586
{
623-
if (std::exchange(first, false))
624-
formula += ' ';
625-
formula += symbol;
626-
if (count > 1)
627-
formula += std::to_string(count);
587+
const auto &[id, name, threeLetterCode, group] =
588+
chem_comp->front().get<std::string, std::string, std::string, std::string>("id", "name", "three_letter_code", "group");
589+
590+
result = construct_compound(db, id, name, threeLetterCode, group);
591+
}
592+
catch (const std::exception &ex)
593+
{
594+
std::throw_with_nested(std::runtime_error("Error loading compound " + id));
628595
}
629596

630-
std::string type;
631-
if (cif::iequals(group, "peptide") or cif::iequals(group, "l-peptide") or cif::iequals(group, "l-peptide linking"))
632-
type = "L-PEPTIDE LINKING";
633-
else if (cif::iequals(group, "dna"))
634-
type = "DNA LINKING";
635-
else if (cif::iequals(group, "rna"))
636-
type = "RNA LINKING";
637-
else
638-
type = "NON-POLYMER";
639-
640-
db["chem_comp"].emplace({
641-
{ "id", id },
642-
{ "name", name },
643-
{ "type", type },
644-
{ "formula", formula },
645-
{ "pdbx_formal_charge", formal_charge },
646-
{ "formula_weight", formula_weight },
647-
{ "three_letter_code", threeLetterCode }
648-
});
649-
650-
m_compounds.push_back(new compound(db));
597+
break;
651598
}
652599
}
653600

654-
// compound *create(const std::string &id) override;
601+
return result;
602+
}
655603

656-
private:
657-
cif::file m_local_file;
658-
};
604+
compound *local_compound_factory_impl::construct_compound(const datablock &rdb, const std::string &id,
605+
const std::string &name, const std::string &three_letter_code, const std::string &group)
606+
{
607+
cif::datablock db(id);
659608

660-
// compound *local_compound_factory_impl::create(const std::string &id)
661-
// {
662-
// compound *result = nullptr;
609+
float formula_weight = 0;
610+
int formal_charge = 0;
611+
std::map<std::string,size_t> formula_data;
663612

664-
// for (auto &db : m_local_file)
665-
// {
666-
// if (db.name() == id)
667-
// {
668-
// cif::datablock db_copy(db);
613+
for (size_t ord = 1; const auto &[atom_id, type_symbol, type, charge, x, y, z] :
614+
rdb["chem_comp_atom"].rows<std::string, std::string, std::string, int, float, float, float>(
615+
"atom_id", "type_symbol", "type", "charge", "x", "y", "z"))
616+
{
617+
auto atom = cif::atom_type_traits(type_symbol);
618+
formula_weight += atom.weight();
669619

670-
// try
671-
// {
672-
// result = new compound(db_copy, 1);
673-
// }
674-
// catch (const std::exception &ex)
675-
// {
676-
// std::throw_with_nested(std::runtime_error("Error loading compound " + id));
677-
// }
620+
formula_data[type_symbol] += 1;
678621

622+
db["chem_comp_atom"].emplace({
623+
{ "comp_id", id },
624+
{ "atom_id", atom_id },
625+
{ "type_symbol", type_symbol },
626+
{ "charge", charge },
627+
{ "model_Cartn_x", x, 3 },
628+
{ "model_Cartn_y", y, 3 },
629+
{ "model_Cartn_z", z, 3 },
630+
{ "pdbx_ordinal", ord++ }
631+
});
679632

680-
// std::shared_lock lock(mMutex);
681-
// m_compounds.push_back(result);
633+
formal_charge += charge;
634+
}
635+
636+
for (size_t ord = 1; const auto &[atom_id_1, atom_id_2, type, aromatic] :
637+
rdb["chem_comp_bond"].rows<std::string, std::string, std::string, bool>("atom_id_1", "atom_id_2", "type", "aromatic"))
638+
{
639+
std::string value_order("SING");
682640

683-
// break;
684-
// }
685-
// }
641+
if (cif::iequals(type, "single") or cif::iequals(type, "sing"))
642+
value_order = "SING";
643+
else if (cif::iequals(type, "double") or cif::iequals(type, "doub"))
644+
value_order = "DOUB";
645+
else if (cif::iequals(type, "triple") or cif::iequals(type, "trip"))
646+
value_order = "TRIP";
686647

687-
// return result;
688-
// }
648+
db["chem_comp_bond"].emplace({
649+
{ "comp_id", id },
650+
{ "atom_id_1", atom_id_1 },
651+
{ "atom_id_2", atom_id_2 },
652+
{ "value_order", value_order },
653+
{ "pdbx_aromatic_flag", aromatic },
654+
// TODO: fetch stereo_config info from chem_comp_chir
655+
{ "pdbx_ordinal", ord++ }
656+
});
657+
}
658+
659+
db.emplace_back(rdb["pdbx_chem_comp_descriptor"]);
660+
661+
std::string formula;
662+
for (bool first = true; const auto &[symbol, count]: formula_data)
663+
{
664+
if (std::exchange(first, false))
665+
formula += ' ';
666+
formula += symbol;
667+
if (count > 1)
668+
formula += std::to_string(count);
669+
}
670+
671+
std::string type;
672+
if (cif::iequals(group, "peptide") or cif::iequals(group, "l-peptide") or cif::iequals(group, "l-peptide linking"))
673+
type = "L-PEPTIDE LINKING";
674+
else if (cif::iequals(group, "dna"))
675+
type = "DNA LINKING";
676+
else if (cif::iequals(group, "rna"))
677+
type = "RNA LINKING";
678+
else
679+
type = "NON-POLYMER";
680+
681+
db["chem_comp"].emplace({
682+
{ "id", id },
683+
{ "name", name },
684+
{ "type", type },
685+
{ "formula", formula },
686+
{ "pdbx_formal_charge", formal_charge },
687+
{ "formula_weight", formula_weight },
688+
{ "three_letter_code", three_letter_code }
689+
});
690+
691+
std::shared_lock lock(mMutex);
692+
693+
auto result = new compound(db);
694+
m_compounds.push_back(result);
695+
return result;
696+
}
689697

690698
// --------------------------------------------------------------------
691699

0 commit comments

Comments
 (0)