From cd8ba0804d3a987f780db9f989bac3ab5ed4930c Mon Sep 17 00:00:00 2001
From: B13nch3n <stefan.bienert@me.com>
Date: Tue, 21 Jul 2020 15:15:46 +0200
Subject: [PATCH] Allow free positioning of entity category in mmCIF files
---
modules/io/src/mol/mmcif_reader.cc | 69 ++++++++++++---------------
modules/io/src/mol/mmcif_reader.hh | 6 +++
modules/io/tests/test_mmcif_reader.cc | 36 +++-----------
3 files changed, 43 insertions(+), 68 deletions(-)
diff --git a/modules/io/src/mol/mmcif_reader.cc b/modules/io/src/mol/mmcif_reader.cc
index 1eb122070..e2f67b119 100644
--- a/modules/io/src/mol/mmcif_reader.cc
+++ b/modules/io/src/mol/mmcif_reader.cc
@@ -660,47 +660,50 @@ void MMCifReader::ParseAndAddAtom(const std::vector<StringRef>& columns)
columns[indices_[GROUP_PDB]][0]=='H');
}
+MMCifReader::MMCifEntityDescMap::iterator MMCifReader::GetEntityDescMapIterator(
+ const String& entity_id)
+{
+ MMCifEntityDescMap::iterator edm_it = entity_desc_map_.find(entity_id);
+ // if the entity ID is not already stored, insert it with empty values
+ if (edm_it == entity_desc_map_.end()) {
+ MMCifEntityDesc desc = {.type=mol::CHAINTYPE_N_CHAINTYPES,
+ .details="",
+ .seqres=""};
+ edm_it = entity_desc_map_.insert(entity_desc_map_.begin(),
+ MMCifEntityDescMap::value_type(entity_id,
+ desc));
+ }
+ return edm_it;
+}
+
void MMCifReader::ParseEntity(const std::vector<StringRef>& columns)
{
- bool store = false; // is it worth storing this record?
- MMCifEntityDesc desc;
+ MMCifEntityDescMap::iterator edm_it =
+ GetEntityDescMapIterator(columns[indices_[E_ID]].str());
// type
if (indices_[E_TYPE] != -1) {
- desc.type = mol::ChainTypeFromString(columns[indices_[E_TYPE]]);
- store = true;
+ // only use the entity type if no other is set, entity_poly type is
+ // more precise, so if that was set before just leave it in
+ if (edm_it->second.type == mol::CHAINTYPE_N_CHAINTYPES) {
+ edm_it->second.type = mol::ChainTypeFromString(columns[indices_[E_TYPE]]);
+ }
+ } else {
+ // don't deal with entities without type
+ entity_desc_map_.erase(edm_it);
+ return;
}
// description
if (indices_[PDBX_DESCRIPTION] != -1) {
- desc.details = columns[indices_[PDBX_DESCRIPTION]].str();
- } else {
- desc.details = "";
- }
-
- if (store) {
- desc.seqres = "";
- entity_desc_map_.insert(
- MMCifEntityDescMap::value_type(columns[indices_[E_ID]].str(),
- desc)
- );
+ edm_it->second.details = columns[indices_[PDBX_DESCRIPTION]].str();
}
}
void MMCifReader::ParseEntityPoly(const std::vector<StringRef>& columns)
{
- // we assume that the entity cat. ALWAYS comes before the entity_poly cat.
- // search entity
MMCifEntityDescMap::iterator edm_it =
- entity_desc_map_.find(columns[indices_[ENTITY_ID]].str());
-
- if (edm_it == entity_desc_map_.end()) {
- throw IOException(this->FormatDiagnostic(STAR_DIAG_ERROR,
- "'entity_poly' category defined before 'entity' for id '" +
- columns[indices_[ENTITY_ID]].str() +
- "' or missing.",
- this->GetCurrentLinenum()));
- }
+ GetEntityDescMapIterator(columns[indices_[ENTITY_ID]].str());
// store type
if (indices_[EP_TYPE] != -1) {
@@ -1713,19 +1716,9 @@ void MMCifReader::ParseStructRefSeqDif(const std::vector<StringRef>& columns)
void MMCifReader::ParsePdbxEntityBranch(const std::vector<StringRef>& columns)
{
- // we assume that the entity cat. ALWAYS comes before the pdbx_entity_branch
- // cat.
- // search entity
+ // get entity/ descreption entry
MMCifEntityDescMap::iterator edm_it =
- entity_desc_map_.find(columns[indices_[BR_ENTITY_ID]].str());
-
- if (edm_it == entity_desc_map_.end()) {
- throw IOException(this->FormatDiagnostic(STAR_DIAG_ERROR,
- "'pdbx_entity_branch' category defined before 'entity' for id '" +
- columns[indices_[BR_ENTITY_ID]].str() +
- "' or missing.",
- this->GetCurrentLinenum()));
- }
+ GetEntityDescMapIterator(columns[indices_[BR_ENTITY_ID]].str());
// store type
if (indices_[BR_ENTITY_TYPE] != -1) {
diff --git a/modules/io/src/mol/mmcif_reader.hh b/modules/io/src/mol/mmcif_reader.hh
index a9339c6ac..03ceb32c9 100644
--- a/modules/io/src/mol/mmcif_reader.hh
+++ b/modules/io/src/mol/mmcif_reader.hh
@@ -629,6 +629,10 @@ private:
} MMCifEntityDesc;
typedef std::map<String, MMCifEntityDesc> MMCifEntityDescMap;
+ /// \brief Get an iterator for MMCifEntityDescMap by finding an element or
+ /// inserting a new one into the map.
+ MMCifEntityDescMap::iterator GetEntityDescMapIterator(const String&);
+
/// \struct assembly information
typedef struct {
String biounit_id; ///< identifier for the bu
@@ -739,3 +743,5 @@ DLLEXPORT_OST_IO String OSTBondOrderToMMCifValueOrder(
}}
#endif
+
+// LocalWords: MMCifEntityDescMap
diff --git a/modules/io/tests/test_mmcif_reader.cc b/modules/io/tests/test_mmcif_reader.cc
index 65673c858..b6fa90f1f 100644
--- a/modules/io/tests/test_mmcif_reader.cc
+++ b/modules/io/tests/test_mmcif_reader.cc
@@ -311,9 +311,13 @@ BOOST_AUTO_TEST_CASE(mmcif_unknown_entity_type)
columns.push_back(StringRef("polymer", 7));
BOOST_CHECK_NO_THROW(tmmcif_p.ParseEntity(columns));
columns.pop_back();
+ columns.pop_back();
+ columns.push_back(StringRef("2", 1));
columns.push_back(StringRef("non-polymer", 11));
BOOST_CHECK_NO_THROW(tmmcif_p.ParseEntity(columns));
columns.pop_back();
+ columns.pop_back();
+ columns.push_back(StringRef("3", 1));
columns.push_back(StringRef("water", 5));
BOOST_CHECK_NO_THROW(tmmcif_p.ParseEntity(columns));
BOOST_TEST_MESSAGE(" done.");
@@ -321,6 +325,8 @@ BOOST_AUTO_TEST_CASE(mmcif_unknown_entity_type)
// negative
BOOST_TEST_MESSAGE(" unknown type...");
columns.pop_back();
+ columns.pop_back();
+ columns.push_back(StringRef("4", 1));
columns.push_back(StringRef("foo", 3));
BOOST_CHECK_THROW(tmmcif_p.ParseEntity(columns), Error);
BOOST_TEST_MESSAGE(" done.");
@@ -404,20 +410,6 @@ BOOST_AUTO_TEST_CASE(mmcif_entity_poly_tests)
seq::SequenceHandle curr = seqres.FindSequence("A");
BOOST_CHECK(curr.GetString() == "VTI");
- BOOST_TEST_MESSAGE(" testing missing corresponding entity entry...");
- {
- mol::EntityHandle eh = mol::CreateEntity();
- std::vector<StringRef> columns;
- TestMMCifReaderProtected tmmcif_p("testfiles/mmcif/atom_site.mmcif", eh);
-
- tmmcif_h.SetCategory(StringRef("entity_poly", 11));
- tmmcif_h.Add(StringRef("entity_id", 9));
- tmmcif_p.OnBeginLoop(tmmcif_h);
-
- columns.push_back(StringRef("1", 1));
- BOOST_CHECK_THROW(tmmcif_p.ParseEntityPoly(columns), IOException);
- }
- BOOST_TEST_MESSAGE(" done.");
BOOST_TEST_MESSAGE(" testing type recognition...");
{
TestMMCifReaderProtected tmmcif_p("testfiles/mmcif/atom_site.mmcif", eh);
@@ -1435,22 +1427,6 @@ BOOST_AUTO_TEST_CASE(mmcif_pdbx_entity_branch_tests)
mmcif_p.Parse();
- BOOST_TEST_MESSAGE(" testing missing corresponding entity entry...");
- {
- mol::EntityHandle eh = mol::CreateEntity();
- std::vector<StringRef> columns;
- TestMMCifReaderProtected tmmcif_p("testfiles/mmcif/atom_site.mmcif", eh);
-
- tmmcif_h.SetCategory(StringRef("pdbx_entity_branch", 18));
- tmmcif_h.Add(StringRef("entity_id", 9));
- tmmcif_h.Add(StringRef("type", 4));
- tmmcif_p.OnBeginLoop(tmmcif_h);
- columns.push_back(StringRef("1", 1));
- columns.push_back(StringRef("oligosaccharide", 15));
-
- BOOST_CHECK_THROW(tmmcif_p.ParsePdbxEntityBranch(columns), IOException);
- }
- BOOST_TEST_MESSAGE(" done.");
BOOST_TEST_MESSAGE(" testing chain type recognition...");
{
TestMMCifReaderProtected tmmcif_p("testfiles/mmcif/atom_site.mmcif", eh);
--
GitLab