From 9c4665d52f9b51d36d60f17502a2a4f575e9bd16 Mon Sep 17 00:00:00 2001
From: Gabriel Studer <gabriel.studer@unibas.ch>
Date: Thu, 4 Jan 2024 13:24:50 +0100
Subject: [PATCH] mmcif writer: delegate chain typing to dedicated functions in
ost mol
---
modules/io/src/mol/mmcif_writer.cc | 81 +-----------------------------
modules/mol/base/src/chain_type.cc | 54 ++++++++++++++++++++
modules/mol/base/src/chain_type.hh | 49 ++++++++++++++++++
3 files changed, 105 insertions(+), 79 deletions(-)
diff --git a/modules/io/src/mol/mmcif_writer.cc b/modules/io/src/mol/mmcif_writer.cc
index be209e559..3cfd364d3 100644
--- a/modules/io/src/mol/mmcif_writer.cc
+++ b/modules/io/src/mol/mmcif_writer.cc
@@ -152,37 +152,6 @@ namespace {
return "other";
}
- String GuessEntityPolyType(ost::mol::ChainType chain_type) {
- // no real guessing but hardcoded response for every polymer chain type in
- // ost::mol::ChainType
-
- // allowed values according to mmcif_pdbx_v50.dic:
- // - cyclic-pseudo-peptide
- // - other
- // - peptide nucleic acid
- // - polydeoxyribonucleotide
- // - polydeoxyribonucleotide/polyribonucleotide hybrid
- // - polypeptide(D)
- // - polypeptide(L)
- // - polyribonucleotide
-
- // added additional type: unknown
- // must be handled by caller
-
- switch(chain_type) {
- case ost::mol::CHAINTYPE_POLY: return "other";
- case ost::mol::CHAINTYPE_POLY_PEPTIDE_D: return "polypeptide(D)";
- case ost::mol::CHAINTYPE_POLY_PEPTIDE_L: return "polypeptide(L)";
- case ost::mol::CHAINTYPE_POLY_DN: return "polydeoxyribonucleotide";
- case ost::mol::CHAINTYPE_POLY_RN: return "polyribonucleotide";
- case ost::mol::CHAINTYPE_POLY_DN_RN: return "polydeoxyribonucleotide/polyribonucleotide hybrid";
- case ost::mol::CHAINTYPE_CYCLIC_PSEUDO_PEPTIDE: return "cyclic-pseudo-peptide";
- case ost::mol::CHAINTYPE_POLY_PEPTIDE_DN_RN: return "peptide nucleic acid";
- case ost::mol::CHAINTYPE_OLIGOSACCHARIDE: return "other";
- default: return "unknown";
- }
- }
-
String GuessEntityType(const ost::mol::ResidueHandleList& res_list) {
// guesses _entity.type based on residue chem classes
@@ -219,41 +188,6 @@ namespace {
return "polymer";
}
- String GuessEntityType(ost::mol::ChainType chain_type) {
- // no real guessing but hardcoded response for every chain type in
- // ost::mol::ChainType
-
- // allowed values according to mmcif_pdbx_v50.dic:
- // - branched
- // - macrolide
- // - non-polymer
- // - polymer
- // - water
-
- // added additional type: unknown
- // must be handled by caller
-
- switch(chain_type) {
- case ost::mol::CHAINTYPE_POLY: return "polymer";
- case ost::mol::CHAINTYPE_NON_POLY: return "non-polymer";
- case ost::mol::CHAINTYPE_WATER: return "water";
- case ost::mol::CHAINTYPE_POLY_PEPTIDE_D: return "polymer";
- case ost::mol::CHAINTYPE_POLY_PEPTIDE_L: return "polymer";
- case ost::mol::CHAINTYPE_POLY_DN: return "polymer";
- case ost::mol::CHAINTYPE_POLY_RN: return "polymer";
- case ost::mol::CHAINTYPE_POLY_SAC_D: return "polymer";
- case ost::mol::CHAINTYPE_POLY_SAC_L: return "polymer";
- case ost::mol::CHAINTYPE_POLY_DN_RN: return "polymer";
- case ost::mol::CHAINTYPE_UNKNOWN: return "unknown";
- case ost::mol::CHAINTYPE_MACROLIDE: return "macrolide";
- case ost::mol::CHAINTYPE_CYCLIC_PSEUDO_PEPTIDE: return "polymer";
- case ost::mol::CHAINTYPE_POLY_PEPTIDE_DN_RN: return "polymer";
- case ost::mol::CHAINTYPE_BRANCHED: return "branched";
- case ost::mol::CHAINTYPE_OLIGOSACCHARIDE: return "branched";
- default: return "unknown";
- }
- }
-
// internal object with all info to fill chem_comp_ category
struct CompInfo {
String type;
@@ -813,22 +747,11 @@ namespace {
std::vector<ost::io::MMCifWriterEntity>& entity_infos) {
// use chain_type info attached to chain to determine
// _entity.type and _entity_poly.type
- String type = GuessEntityType(chain_type);
- if(type == "unknown") {
- std::stringstream ss;
- ss << "Each chain must have valid chain type set, got " << chain_type;
- throw ost::io::IOException(ss.str());
- }
+ String type = ost::mol::EntityTypeFromChainType(chain_type);
bool is_poly = type == "polymer";
String poly_type = "";
if(is_poly) {
- poly_type = GuessEntityPolyType(chain_type);
- if(poly_type == "unknown") {
- std::stringstream ss;
- ss << "Each polymer chain must have valid polymer chain type set, got ";
- ss << chain_type;
- throw ost::io::IOException(ss.str());
- }
+ poly_type = ost::mol::EntityPolyTypeFromChainType(chain_type);
}
return SetupEntity(asym_chain_name, type, poly_type, res_list,
resnum_alignment, entity_infos);
diff --git a/modules/mol/base/src/chain_type.cc b/modules/mol/base/src/chain_type.cc
index 7604bbc50..39be8389c 100644
--- a/modules/mol/base/src/chain_type.cc
+++ b/modules/mol/base/src/chain_type.cc
@@ -121,4 +121,58 @@ String StringFromChainType(ChainType type)
throw Error(ss.str());
}
+String EntityTypeFromChainType(ChainType type) {
+ switch(type) {
+ case ost::mol::CHAINTYPE_POLY: return "polymer";
+ case ost::mol::CHAINTYPE_NON_POLY: return "non-polymer";
+ case ost::mol::CHAINTYPE_WATER: return "water";
+ case ost::mol::CHAINTYPE_POLY_PEPTIDE_D: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_PEPTIDE_L: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_DN: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_RN: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_SAC_D: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_SAC_L: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_DN_RN: return "polymer";
+ case ost::mol::CHAINTYPE_MACROLIDE: return "macrolide";
+ case ost::mol::CHAINTYPE_CYCLIC_PSEUDO_PEPTIDE: return "polymer";
+ case ost::mol::CHAINTYPE_POLY_PEPTIDE_DN_RN: return "polymer";
+ case ost::mol::CHAINTYPE_BRANCHED: return "branched";
+ case ost::mol::CHAINTYPE_OLIGOSACCHARIDE: return "branched";
+ }
+
+ std::stringstream ss("Unknown ChainType item found: '");
+ ss << type << "'!";
+ throw Error(ss.str());
+}
+
+String EntityPolyTypeFromChainType(ChainType type) {
+ switch(type) {
+ case ost::mol::CHAINTYPE_POLY: return "other";
+ case ost::mol::CHAINTYPE_POLY_PEPTIDE_D: return "polypeptide(D)";
+ case ost::mol::CHAINTYPE_POLY_PEPTIDE_L: return "polypeptide(L)";
+ case ost::mol::CHAINTYPE_POLY_DN: return "polydeoxyribonucleotide";
+ case ost::mol::CHAINTYPE_POLY_RN: return "polyribonucleotide";
+ case ost::mol::CHAINTYPE_POLY_SAC_D: return "other"; // older dictionaries have "polysaccharide(D)"
+ case ost::mol::CHAINTYPE_POLY_SAC_L: return "other"; // older dictionaries have "polysaccharide(L)"
+ case ost::mol::CHAINTYPE_POLY_DN_RN: return "polydeoxyribonucleotide/polyribonucleotide hybrid";
+ case ost::mol::CHAINTYPE_CYCLIC_PSEUDO_PEPTIDE: return "cyclic-pseudo-peptide";
+ case ost::mol::CHAINTYPE_POLY_PEPTIDE_DN_RN: return "peptide nucleic acid";
+ }
+
+ std::stringstream ss("Cannot return entity poly type from chain of type: '");
+ ss << type << "'!";
+ throw Error(ss.str());
+}
+
+String BranchedTypeFromChainType(ChainType type) {
+ switch(type) {
+ case ost::mol::CHAINTYPE_BRANCHED: return "oligosaccharide"; // the only one
+ case ost::mol::CHAINTYPE_OLIGOSACCHARIDE: return "oligosaccharide";
+ }
+
+ std::stringstream ss("Cannot return branched type from chain of type: '");
+ ss << type << "'!";
+ throw Error(ss.str());
+}
+
}} //ns
diff --git a/modules/mol/base/src/chain_type.hh b/modules/mol/base/src/chain_type.hh
index 839844827..6420f0143 100644
--- a/modules/mol/base/src/chain_type.hh
+++ b/modules/mol/base/src/chain_type.hh
@@ -79,6 +79,55 @@ ChainType DLLEXPORT_OST_MOL ChainTypeFromString(const String& identifier);
/// unknown type
String DLLEXPORT_OST_MOL StringFromChainType(ChainType type);
+/// \brief Return _entity.type consistent with respective mmCIF vocabulary
+/// (mmcif_pdbx_v50):
+/// - branched
+/// - macrolide
+/// - non-polymer
+/// - polymer
+/// - water
+///
+/// For consistency with older vocabularies, CHAINTYPE_POLY_SAC_D
+/// and CHAINTYPE_POLY_SAC_L return "polymer"
+///
+/// \param type ChainType to be translated
+///
+/// \return String corresponding to the input, throws a ost::Error on
+/// unknown type
+String DLLEXPORT_OST_MOL EntityTypeFromChainType(ChainType type);
+
+/// \brief Return _entity_poly.type consistent with mmCIF dictionary
+/// (mmcif_pdbx_v50):
+/// - cyclic-pseudo-peptide
+/// - other
+/// - peptide nucleic acid
+/// - polydeoxyribonucleotide
+/// - polydeoxyribonucleotide/polyribonucleotide hybrid
+/// - polypeptide(D)
+/// - polypeptide(L)
+/// - polyribonucleotide
+///
+/// For consistency with older dictionaries, CHAINTYPE_POLY_SAC_D
+/// and CHAINTYPE_POLY_SAC_L are still accepted but return "other".
+/// Older dictionaries still had "polysaccharide(D)" and
+/// "polysaccharide(L)""
+///
+/// \param type ChainType to be translated
+///
+/// \return String corresponding to the input, throws a ost::Error on
+/// unknown type or if it's not of _entity.type polymer
+String DLLEXPORT_OST_MOL EntityPolyTypeFromChainType(ChainType type);
+
+/// \brief Return pdbx_entity_branch.type consistent with mmCIF dictionary
+/// (mmcif_pdbx_v50):
+/// - oligosaccharide
+///
+/// \param type ChainType to be translated
+///
+/// \return String corresponding to the input, throws a ost::Error on
+/// unknown type or if it's not of _entity.type branched
+String DLLEXPORT_OST_MOL BranchedTypeFromChainType(ChainType type);
+
}} //ns
#endif
--
GitLab