From ca5d8dd46e0f53dad344e45b3b6240a21f63b47b Mon Sep 17 00:00:00 2001
From: Stefan Bienert <stefan.bienert@unibas.ch>
Date: Mon, 3 Jun 2013 11:06:54 +0200
Subject: [PATCH] Added pdbx_struct_assembly.method_details to mmCIF parser
---
modules/io/doc/mmcif.rst | 19 +++++++++++++++++--
modules/io/pymod/export_mmcif_io.cc | 4 ++++
modules/io/src/mol/mmcif_info.hh | 12 ++++++++++++
modules/io/src/mol/mmcif_reader.cc | 26 +++++++++++++++++---------
modules/io/src/mol/mmcif_reader.hh | 11 +++++++++--
modules/io/tests/test_io_mmcif.py | 4 ++++
modules/io/tests/test_mmcif_info.cc | 2 ++
modules/io/tests/test_mmcif_reader.cc | 1 +
8 files changed, 66 insertions(+), 13 deletions(-)
diff --git a/modules/io/doc/mmcif.rst b/modules/io/doc/mmcif.rst
index b8c63d9de..0a6520a47 100644
--- a/modules/io/doc/mmcif.rst
+++ b/modules/io/doc/mmcif.rst
@@ -501,6 +501,13 @@ of the annotation available.
Also available as :meth:`GetDetails`. May also be modified by
:meth:`SetDetails`.
+ .. attribute:: method_details
+
+ Details about the method used to determine this biological assembly.
+
+ Also available as :meth:`GetMethodDetails`. May also be modified by
+ :meth:`SetMethodDetails`.
+
.. attribute:: chains
Chains involved in this bio unit. If not provided, resembles an empty list.
@@ -552,6 +559,14 @@ of the annotation available.
See :attr:`details`
+ .. method:: GetMethodDetails()
+
+ See :attr:`method_details`
+
+ .. method:: SetMethodDetails(details)
+
+ See :attr:`method_details`
+
.. method:: GetChainList()
See :attr:`chains`
@@ -997,5 +1012,5 @@ of the annotation available.
.. LocalWords: SetChainList MMCifInfoTransOp ChainTypes MMCifInfoStructRef
.. LocalWords: MMCifInfoRevisions bool difs MMCifInfoStructRefSeqDif rnum
.. LocalWords: SetDateOriginal GetDateOriginal yyyy operationsintervalls
-.. LocalWords: chainintervalls GetChainIntervalList
-.. LocalWords: GetOperationsIntervalList
+.. LocalWords: chainintervalls GetChainIntervalList GetMethodDetails
+.. LocalWords: GetOperationsIntervalList SetMethodDetails
diff --git a/modules/io/pymod/export_mmcif_io.cc b/modules/io/pymod/export_mmcif_io.cc
index 9a21bdb1a..d32133571 100644
--- a/modules/io/pymod/export_mmcif_io.cc
+++ b/modules/io/pymod/export_mmcif_io.cc
@@ -184,6 +184,8 @@ void export_mmcif_io()
class_<MMCifInfoBioUnit>("MMCifInfoBioUnit", init<>())
.def("SetDetails", &MMCifInfoBioUnit::SetDetails)
.def("GetDetails", &MMCifInfoBioUnit::GetDetails)
+ .def("SetMethodDetails", &MMCifInfoBioUnit::SetMethodDetails)
+ .def("GetMethodDetails", &MMCifInfoBioUnit::GetMethodDetails)
.def("AddChain", &MMCifInfoBioUnit::AddChain)
.def("SetChainList", &MMCifInfoBioUnit::SetChainList)
.def("GetChainList", make_function(&MMCifInfoBioUnit::GetChainList,
@@ -201,6 +203,8 @@ void export_mmcif_io()
.def("GetID", &MMCifInfoBioUnit::GetID)
.add_property("details", &MMCifInfoBioUnit::GetDetails,
&MMCifInfoBioUnit::SetDetails)
+ .add_property("method_details", &MMCifInfoBioUnit::GetMethodDetails,
+ &MMCifInfoBioUnit::SetMethodDetails)
.add_property("chains", make_function(&MMCifInfoBioUnit::GetChainList,
return_value_policy<copy_const_reference>()))
.add_property("chainintervalls", make_function(
diff --git a/modules/io/src/mol/mmcif_info.hh b/modules/io/src/mol/mmcif_info.hh
index 52a44aedc..593fed6f2 100644
--- a/modules/io/src/mol/mmcif_info.hh
+++ b/modules/io/src/mol/mmcif_info.hh
@@ -264,6 +264,17 @@ public:
/// \return details
String GetDetails() const { return details_; }
+ /// \brief Set method details
+ ///
+ /// \param method details
+ void SetMethodDetails(String method_details) {
+ method_details_ = method_details;
+ }
+ /// \brief Get method details
+ ///
+ /// \return method details
+ String GetMethodDetails() const { return method_details_; }
+
/// \brief Add a chain name
///
/// \param chain chain name
@@ -360,6 +371,7 @@ public:
private:
String id_; ///< pdbx_struct_assembly.id
String details_; ///< pdbx_struct_assembly.details
+ String method_details_; ///< pdbx_struct_assembly.method_details
std::vector<String> chains_; ///< all chains of this this assembly
std::vector<std::pair<int, int> > tr_chains_; //< chains of a transformation
std::vector<std::vector<MMCifInfoTransOpPtr> > operations_;
diff --git a/modules/io/src/mol/mmcif_reader.cc b/modules/io/src/mol/mmcif_reader.cc
index cd21e62a1..52cd7e1ae 100644
--- a/modules/io/src/mol/mmcif_reader.cc
+++ b/modules/io/src/mol/mmcif_reader.cc
@@ -224,7 +224,7 @@ bool MMCifReader::OnBeginLoop(const StarLoopDesc& header)
this->TryStoreIdx(PSA_ID, "id", header);
// optional
indices_[PSA_DETAILS] = header.GetIndex("details");
- //indices_[METHOD_DETAILS] = header.GetIndex("method_details");
+ indices_[METHOD_DETAILS] = header.GetIndex("method_details");
cat_available = true;
} else if (header.GetCategory() == "pdbx_struct_assembly_gen") {
category_ = PDBX_STRUCT_ASSEMBLY_GEN;
@@ -870,15 +870,22 @@ void MMCifReader::ParseRefine(const std::vector<StringRef>& columns)
void MMCifReader::ParsePdbxStructAssembly(const std::vector<StringRef>& columns)
{
+ MMCifPSAEntry psa;
+
if (indices_[PSA_DETAILS] != -1) {
- bu_origin_map_.insert(std::pair<String,
- String>(columns[indices_[PSA_ID]].str(),
- columns[indices_[PSA_DETAILS]].str()));
+ psa.details = columns[indices_[PSA_DETAILS]].str();
} else {
- bu_origin_map_.insert(std::pair<String,
- String>(columns[indices_[PSA_ID]].str(),
- "?"));
+ psa.details = "?";
}
+
+ if (indices_[METHOD_DETAILS] != -1) {
+ psa.method_details = columns[indices_[METHOD_DETAILS]].str();
+ } else {
+ psa.method_details = "?";
+ }
+
+ bu_origin_map_.insert(std::pair<String,
+ MMCifPSAEntry>(columns[indices_[PSA_ID]].str(), psa));
}
void MMCifReader::StoreExpression(const char* l, const char* s,
@@ -1614,7 +1621,7 @@ void MMCifReader::OnEndData()
std::vector<std::vector<String> >::const_iterator aol_it;
std::vector<String>::const_iterator aob_it;
std::vector<MMCifInfoTransOpPtr> operation_list;
- std::map<String, String>::const_iterator buom_it;
+ std::map<String, MMCifPSAEntry>::const_iterator buom_it;
std::vector<MMCifInfoTransOpPtr> operations = info_.GetOperations();
info_.SetStructRefs(struct_refs_);
std::vector<MMCifInfoTransOpPtr>::const_iterator buop_it;
@@ -1631,7 +1638,8 @@ void MMCifReader::OnEndData()
bua_it->biounit_id +
"' found as requested by pdbx_struct_assembly_gen."));
}
- biounit.SetDetails(buom_it->second);
+ biounit.SetDetails(buom_it->second.details);
+ biounit.SetMethodDetails(buom_it->second.method_details);
biounit.SetID(buom_it->first);
biounit.SetChainList(bua_it->chains);
diff --git a/modules/io/src/mol/mmcif_reader.hh b/modules/io/src/mol/mmcif_reader.hh
index d91cf3403..15cc727ad 100644
--- a/modules/io/src/mol/mmcif_reader.hh
+++ b/modules/io/src/mol/mmcif_reader.hh
@@ -578,6 +578,13 @@ private:
} MMCifHSEntry;
typedef std::vector<MMCifHSEntry> MMCifHSVector;
+ /// \struct store pdbx_struct_assembly info
+ typedef struct {
+ String details;
+ String method_details;
+ } MMCifPSAEntry;
+ typedef std::map<String, MMCifPSAEntry> MMCifPSAMap;
+
// members
MMCifCategory category_;
int category_counts_[DONT_KNOW+1]; ///< overall no. of atom_site loops
@@ -605,10 +612,10 @@ private:
MMCifInfo info_; ///< info container
MMCifCitationAuthorMap authors_map_;
MMCifBioUAssemblyVector bu_assemblies_;
- std::map<String, String> bu_origin_map_; ///< pdbx_struct_assembly.details
+ MMCifPSAMap bu_origin_map_; ///< pdbx_struct_assembly
MMCifHSVector helix_list_; ///< for storing struct_conf sec.struct. data
MMCifHSVector strand_list_; ///< for storing struct_conf sec.struct. data
- MMCifInfoStructRefs struct_refs_;
+ MMCifInfoStructRefs struct_refs_;
};
}}
diff --git a/modules/io/tests/test_io_mmcif.py b/modules/io/tests/test_io_mmcif.py
index 4b64aa554..3c1a0ec23 100644
--- a/modules/io/tests/test_io_mmcif.py
+++ b/modules/io/tests/test_io_mmcif.py
@@ -68,7 +68,11 @@ class TestMMCifInfo(unittest.TestCase):
def test_mmcifinfo_biounit(self):
b = io.MMCifInfoBioUnit()
b.SetDetails('Details')
+ b.SetMethodDetails('MethodDetails')
self.assertEquals(b.GetDetails(), 'Details')
+ self.assertEquals(b.GetMethodDetails(), 'MethodDetails')
+ b.method_details = 'AttrMethodDetails'
+ self.assertEquals(b.method_details, 'AttrMethodDetails')
b.AddChain('A')
b.AddChain('B')
cl = b.GetChainList()
diff --git a/modules/io/tests/test_mmcif_info.cc b/modules/io/tests/test_mmcif_info.cc
index 4b204a2a5..69f369a4d 100644
--- a/modules/io/tests/test_mmcif_info.cc
+++ b/modules/io/tests/test_mmcif_info.cc
@@ -109,11 +109,13 @@ BOOST_AUTO_TEST_CASE(mmcif_info_biounit)
MMCifInfoBioUnit bu = MMCifInfoBioUnit();
bu.SetDetails("author_defined_assembly");
+ bu.SetMethodDetails("PISA");
bu.SetID("1");
bu.AddChain("B");
bu.AddChain("A");
BOOST_CHECK(bu.GetDetails() == "author_defined_assembly");
+ BOOST_CHECK(bu.GetMethodDetails() == "PISA");
BOOST_CHECK(bu.GetChainList().back() == "A");
std::vector<std::pair<int, int> > tr = bu.GetChainIntervalList();
BOOST_CHECK(tr[0].first == 0);
diff --git a/modules/io/tests/test_mmcif_reader.cc b/modules/io/tests/test_mmcif_reader.cc
index 0b9e94b58..6f6866622 100644
--- a/modules/io/tests/test_mmcif_reader.cc
+++ b/modules/io/tests/test_mmcif_reader.cc
@@ -1266,6 +1266,7 @@ BOOST_AUTO_TEST_CASE(mmcif_testreader)
BOOST_CHECK(mmcif_p.GetInfo().GetMethod().str() == "Deep-fry");
BOOST_CHECK(mmcif_p.GetInfo().GetBioUnits().back().GetDetails() ==
"author_defined_assembly");
+ BOOST_CHECK(mmcif_p.GetInfo().GetBioUnits().back().GetMethodDetails() == "?");
BOOST_CHECK(mmcif_p.GetInfo().GetBioUnits().back().GetID() == "2");
BOOST_CHECK(mmcif_p.GetInfo().GetBioUnits().back().GetChainList().back() ==
"F");
--
GitLab