From cd7013478638b9330113b0897dbc6373dc228c7a Mon Sep 17 00:00:00 2001
From: Gabriel Studer <gabriel.studer@unibas.ch>
Date: Tue, 30 May 2023 12:17:32 +0200
Subject: [PATCH] read SEQRES by default in mmcif reader
Performance wise this doesn't really make a difference but properly
setting the IsLigand property depends on it.
---
modules/io/pymod/__init__.py | 6 +++---
modules/io/src/mol/mmcif_reader.cc | 20 +++++++++++++-------
modules/io/src/mol/mmcif_reader.hh | 2 +-
modules/io/tests/test_mmcif_reader.cc | 1 +
4 files changed, 18 insertions(+), 11 deletions(-)
diff --git a/modules/io/pymod/__init__.py b/modules/io/pymod/__init__.py
index 0f8131382..ca619a156 100644
--- a/modules/io/pymod/__init__.py
+++ b/modules/io/pymod/__init__.py
@@ -393,14 +393,15 @@ def LoadMMCIF(filename, fault_tolerant=None, calpha_only=None,
interpreted as the pdb id.
:type remote: :class:`bool`
- :param seqres: Whether to read SEQRES records. If True, a
+ :param seqres: Whether to return SEQRES records. If True, a
:class:`~ost.seq.SequenceList` object is returned as the second
item. The sequences in the list are named according to the
mmCIF chain name.
This feature requires a default
:class:`compound library <ost.conop.CompoundLib>`
to be defined and accessible via
- :func:`~ost.conop.GetDefaultLib` or an empty list is returned.
+ :func:`~ost.conop.GetDefaultLib`. One letter codes of non
+ standard compounds are set to X otherwise.
:type seqres: :class:`bool`
:param info: Whether to return an info container with the other output.
@@ -434,7 +435,6 @@ def LoadMMCIF(filename, fault_tolerant=None, calpha_only=None,
try:
ent = mol.CreateEntity()
reader = MMCifReader(filename, ent, prof)
- reader.read_seqres = seqres
# NOTE: to speed up things, we could introduce a restrict_chains parameter
# similar to the one in LoadPDB. Here, it would have to be a list/set
diff --git a/modules/io/src/mol/mmcif_reader.cc b/modules/io/src/mol/mmcif_reader.cc
index e4ab24090..7f7966ce1 100644
--- a/modules/io/src/mol/mmcif_reader.cc
+++ b/modules/io/src/mol/mmcif_reader.cc
@@ -24,9 +24,14 @@
#include <ost/dyn_cast.hh>
#include <ost/mol/xcs_editor.hh>
#include <ost/conop/conop.hh>
+#include <ost/conop/minimal_compound_lib.hh>
#include <ost/io/mol/mmcif_reader.hh>
+
+
+#include <iostream>
+
namespace ost { namespace io {
@@ -65,7 +70,7 @@ void MMCifReader::Init()
curr_chain_ = mol::ChainHandle();
curr_residue_ = mol::ResidueHandle();
seqres_ = seq::CreateSequenceList();
- read_seqres_ = false;
+ read_seqres_ = true;
warned_rule_based_ = false;
info_ = MMCifInfo();
}
@@ -742,15 +747,16 @@ void MMCifReader::ParseEntityPoly(const std::vector<StringRef>& columns)
} else if (indices_[PDBX_SEQ_ONE_LETTER_CODE] != -1) {
seqres=columns[indices_[PDBX_SEQ_ONE_LETTER_CODE]];
- conop::CompoundLibPtr comp_lib=conop::Conopology::Instance()
- .GetDefaultLib();
+ conop::CompoundLibBasePtr comp_lib=conop::Conopology::Instance()
+ .GetDefaultLib();
if (!comp_lib) {
if (!warned_rule_based_) {
- LOG_WARNING("SEQRES import requires a compound library. "
- "Ignoring SEQRES records");
+ LOG_WARNING("SEQRES import requires a valid compound library to "
+ "handle non standard compounds. Their One letter "
+ "codes will be set to X.");
}
warned_rule_based_=true;
- return;
+ comp_lib = conop::CompoundLibBasePtr(new ost::conop::MinimalCompoundLib);
}
edm_it->second.seqres = this->ConvertSEQRES(seqres.str_no_whitespace(),
comp_lib);
@@ -763,7 +769,7 @@ void MMCifReader::ParseEntityPoly(const std::vector<StringRef>& columns)
}
String MMCifReader::ConvertSEQRES(const String& seqres,
- conop::CompoundLibPtr comp_lib)
+ conop::CompoundLibBasePtr comp_lib)
{
String can_seqres;
for (String::const_iterator i=seqres.begin(), e=seqres.end(); i!=e; ++i) {
diff --git a/modules/io/src/mol/mmcif_reader.hh b/modules/io/src/mol/mmcif_reader.hh
index 3d10f03d4..1f21c3928 100644
--- a/modules/io/src/mol/mmcif_reader.hh
+++ b/modules/io/src/mol/mmcif_reader.hh
@@ -240,7 +240,7 @@ protected:
/// not possible, however, since the PDB assigns multiple one letter codes
/// to some of the residues. To be consistent, we have to do the conversion
/// on our own.
- String ConvertSEQRES(const String& seqres, conop::CompoundLibPtr compound_lib);
+ String ConvertSEQRES(const String& seqres, conop::CompoundLibBasePtr compound_lib);
/// \brief Fetch mmCIF citation_author information
///
/// \param columns data row
diff --git a/modules/io/tests/test_mmcif_reader.cc b/modules/io/tests/test_mmcif_reader.cc
index bee89e0e2..e03eb0073 100644
--- a/modules/io/tests/test_mmcif_reader.cc
+++ b/modules/io/tests/test_mmcif_reader.cc
@@ -414,6 +414,7 @@ BOOST_AUTO_TEST_CASE(mmcif_entity_poly_tests)
BOOST_TEST_MESSAGE(" testing type recognition...");
{
TestMMCifReaderProtected tmmcif_p("testfiles/mmcif/atom_site.mmcif", eh);
+ tmmcif_p.SetReadSeqRes(false);
std::vector<StringRef> columns;
// create corresponding entity entry
--
GitLab