NCBI C++ ToolKit
|
#include <ncbi_pch.hpp>
#include <objects/general/Dbtag.hpp>
#include <objects/general/Object_id.hpp>
#include <objects/seq/Seq_annot.hpp>
#include <objects/seq/Seq_data.hpp>
#include <objects/seq/Seq_descr.hpp>
#include <objects/seq/Seq_inst.hpp>
#include <objects/seq/Seqdesc.hpp>
#include <objects/seq/NCBIeaa.hpp>
#include <objects/seq/IUPACaa.hpp>
#include <objects/seq/NCBIstdaa.hpp>
#include <objects/seqfeat/BioSource.hpp>
#include <objects/seqfeat/Org_ref.hpp>
#include <objects/seqloc/PDB_seq_id.hpp>
#include <objects/seqblock/PDB_block.hpp>
#include <objects/seqloc/Giimport_id.hpp>
#include <objects/seqloc/Patent_seq_id.hpp>
#include <objects/seqset/Bioseq_set.hpp>
#include <objects/id1/id1_client.hpp>
#include <util/sequtil/sequtil_convert.hpp>
#include <algo/structure/cd_utils/cuSequence.hpp>
#include <algo/structure/cd_utils/cuDbPriority.hpp>
Go to the source code of this file.
Go to the SVN repository for this file.
Definition at line 704 of file cuSequence.cpp.
References CBioseq_Base::IsSetDescr(), result, CSeq_descr_Base::Set(), CSeqdesc_Base::SetComment(), and CBioseq_Base::SetDescr().
bool CopyBioseqWithType | ( | const CRef< CSeq_entry > & | seqEntry, |
CSeq_id::E_Choice | choice, | ||
CRef< CBioseq > & | seqEntryBioseq | ||
) |
Definition at line 640 of file cuSequence.cpp.
References CSerialObject::Assign(), CopySeqIdsOfType(), CSeq_entry_Base::GetSeq(), CBioseq_set_Base::GetSeq_set(), CSeq_entry_Base::GetSet(), CSeq_entry_Base::IsSeq(), CSeq_entry_Base::IsSet(), CRef< C, Locker >::NotEmpty(), and result.
Definition at line 487 of file cuSequence.cpp.
References CSeq_id::Assign(), CBioseq_Base::GetId(), and result.
Referenced by ExtractGi().
Definition at line 522 of file cuSequence.cpp.
References CSeq_id::Assign(), CBioseq_Base::GetId(), and result.
Referenced by ExtractPdbMolChain().
unsigned int CopySeqIdsOfType | ( | const CBioseq & | bioseq, |
CSeq_id::E_Choice | choice, | ||
list< CRef< CSeq_id > > & | idsOfType | ||
) |
Definition at line 595 of file cuSequence.cpp.
References CBioseq_Base::GetId().
Referenced by CopyBioseqWithType(), CopySeqIdsOfType(), and GetBioseqWithType().
unsigned int CopySeqIdsOfType | ( | const CRef< CSeq_entry > & | seqEntry, |
CSeq_id::E_Choice | choice, | ||
list< CRef< CSeq_id > > & | idsOfType | ||
) |
Definition at line 610 of file cuSequence.cpp.
References CopySeqIdsOfType(), CSeq_entry_Base::GetSeq(), CBioseq_set_Base::GetSeq_set(), CSeq_entry_Base::GetSet(), CSeq_entry_Base::IsSeq(), CSeq_entry_Base::IsSet(), and CRef< C, Locker >::NotEmpty().
bool extractBioseqInfo | ( | const CRef< CBioseq > | bioseq, |
BioseqInfo & | info | ||
) |
Definition at line 889 of file cuSequence.cpp.
References CTextseq_id_Base::CanGetAccession(), CTextseq_id_Base::CanGetVersion(), CSeq_descr_Base::Get(), CTextseq_id_Base::GetAccession(), CBioseq_Base::GetDescr(), CBioseq_Base::GetId(), CTextseq_id_Base::GetVersion(), info, and CBioseq_Base::IsSetDescr().
Definition at line 508 of file cuSequence.cpp.
References CopyGiSeqId(), CSeq_id_Base::GetGi(), result, and ZERO_GI.
bool ExtractPdbMolChain | ( | const CRef< CBioseq > & | bioseq, |
string & | pdbMol, | ||
string & | pdbChain, | ||
unsigned int | nth | ||
) |
Definition at line 544 of file cuSequence.cpp.
References CopyPdbSeqId(), CAliasBase< TPrim >::Get(), CPDB_seq_id::GetEffectiveChain_id(), CPDB_seq_id_Base::GetMol(), CSeq_id_Base::GetPdb(), and result.
bool GetAccAndVersion | ( | const CRef< CBioseq > | bioseq, |
string & | acc, | ||
int & | version, | ||
CRef< CSeq_id > & | seqId | ||
) |
Definition at line 420 of file cuSequence.cpp.
References CSeq_id::Assign(), CTextseq_id_Base::CanGetAccession(), CTextseq_id_Base::CanGetVersion(), CTextseq_id_Base::GetAccession(), CBioseq_Base::GetId(), CTextseq_id_Base::GetVersion(), and dtl::version.
Referenced by CDRefresher::addSequence(), CDUpdater::findRowsWithOldSeq(), CDRefresher::hasOlderVersion(), and CDRefresher::refresh().
void GetAccessionAndDatabaseSource | ( | const CRef< CSeq_id > & | seqID, |
string & | accession, | ||
string & | dbSource, | ||
bool | getGenericSource | ||
) |
Definition at line 811 of file cuSequence.cpp.
References CTextseq_id_Base::CanGetAccession(), CGiimport_id_Base::CanGetDb(), CTextseq_id_Base::CanGetName(), CCdDbPriority::eDPUnknown, CAliasBase< TPrim >::Get(), CTextseq_id_Base::GetAccession(), CDbtag_Base::GetDb(), CGiimport_id_Base::GetDb(), CPDB_seq_id::GetEffectiveChain_id(), CSeq_id_Base::GetGeneral(), CSeq_id_Base::GetGi(), CSeq_id_Base::GetGibbmt(), CSeq_id_Base::GetGibbsq(), CSeq_id_Base::GetGiim(), CObject_id_Base::GetId(), CGiimport_id_Base::GetId(), CSeq_id_Base::GetLocal(), CPDB_seq_id_Base::GetMol(), CTextseq_id_Base::GetName(), CSeq_id_Base::GetPatent(), CSeq_id_Base::GetPdb(), CPatent_seq_id_Base::GetSeqid(), CCdDbPriority::GetSourceName(), CObject_id_Base::GetStr(), CDbtag_Base::GetTag(), CSeq_id::GetTextseq_Id(), NStr::IntToString(), CSeq_id_Base::IsGeneral(), CSeq_id_Base::IsGi(), CSeq_id_Base::IsGibbmt(), CSeq_id_Base::IsGibbsq(), CSeq_id_Base::IsGiim(), CObject_id_Base::IsId(), CSeq_id_Base::IsLocal(), CSeq_id_Base::IsPatent(), CSeq_id_Base::IsPdb(), CDbtag_Base::IsSetDb(), CDbtag_Base::IsSetTag(), CObject_id_Base::IsStr(), NStr::NumericToString(), CCdDbPriority::SeqIdTypeToSource(), and CSeq_id_Base::Which().
Referenced by GetAccessionForSeqId(), and GetDbSourceForSeqId().
Definition at line 803 of file cuSequence.cpp.
References GetAccessionAndDatabaseSource().
bool GetBioseqWithType | ( | CRef< CSeq_entry > & | seqEntry, |
CSeq_id::E_Choice | choice, | ||
CRef< CBioseq > & | seqEntryBioseq | ||
) |
Definition at line 672 of file cuSequence.cpp.
References CopySeqIdsOfType(), CSeq_entry_Base::GetSeq(), CSeq_entry_Base::IsSeq(), CSeq_entry_Base::IsSet(), CRef< C, Locker >::NotEmpty(), result, CSeq_entry_Base::SetSeq(), CBioseq_set_Base::SetSeq_set(), and CSeq_entry_Base::SetSet().
Definition at line 97 of file cuSequence.cpp.
References database, CDbtag_Base::GetDb(), CSeq_id_Base::GetGeneral(), CDbtag_Base::GetTag(), CSeq_id_Base::IsGeneral(), and CObject_id_Base::IsId().
Referenced by GetPssmIdFromSeqAlign().
Definition at line 796 of file cuSequence.cpp.
References GetAccessionAndDatabaseSource().
Definition at line 112 of file cuSequence.cpp.
References CBioseq_Base::GetAnnot(), and CBioseq_Base::IsSetAnnot().
Referenced by CCdCore::GetMmdbId(), and AlignmentCollection::IsStruct().
Definition at line 298 of file cuSequence.cpp.
References CAliasBase< TPrim >::Get(), CBioseq_Base::GetInst(), CSeq_data_Base::GetIupacaa(), CSeq_data_Base::GetNcbieaa(), CSeq_data_Base::GetNcbistdaa(), CSeq_inst_Base::GetSeq_data(), CSeq_data_Base::IsIupacaa(), CSeq_data_Base::IsNcbieaa(), CSeq_data_Base::IsNcbistdaa(), CSeq_inst_Base::IsSetSeq_data(), and NcbistdaaToNcbieaaString().
Referenced by AlignmentCollection::GetSequenceForRow(), CCdCore::GetSequenceStringByIndex(), CDRefresher::refresh(), and CdBlaster::truncateBioseq().
bool GetNcbieaaString | ( | const CRef< CSeq_entry > & | Seq, |
string & | Str | ||
) |
Definition at line 289 of file cuSequence.cpp.
References CBioseq_Base::GetInst(), CSeq_entry_Base::GetSeq(), CSeq_entry_Base::IsSeq(), and CSeq_inst_Base::IsSetSeq_data().
Referenced by GetResidueAtPosition().
Definition at line 315 of file cuSequence.cpp.
References CSeqConvert::Convert(), CSeqUtil::e_Iupacaa, CSeqUtil::e_Ncbieaa, CSeqUtil::e_Ncbistdaa, CAliasBase< TPrim >::Get(), CBioseq_Base::GetInst(), CSeq_data_Base::GetIupacaa(), CSeq_data_Base::GetNcbieaa(), CSeq_data_Base::GetNcbistdaa(), CSeq_inst_Base::GetSeq_data(), CSeq_data_Base::IsIupacaa(), CSeq_data_Base::IsNcbieaa(), CSeq_data_Base::IsNcbistdaa(), and CSeq_inst_Base::IsSetSeq_data().
Referenced by PssmScorer::score().
bool GetPDBBlockFromSeqEntry | ( | CRef< CSeq_entry > | seqEntry, |
CRef< CPDB_block > & | pdbBlock | ||
) |
Definition at line 446 of file cuSequence.cpp.
References CSeq_descr_Base::Get(), CBioseq_Base::GetDescr(), CBioseq_set_Base::GetDescr(), CSeq_entry_Base::GetSeq(), CBioseq_set_Base::GetSeq_set(), CSeq_entry_Base::GetSet(), CSeq_entry_Base::IsSeq(), CRef< C, Locker >::Reset(), and CSeqdesc_Base::SetPdb().
Referenced by CDUpdater::reformatBioseq().
Definition at line 349 of file cuSequence.cpp.
References CAliasBase< TPrim >::Get(), CBioseq_Base::GetInst(), CSeq_data_Base::GetIupacaa(), CSeq_data_Base::GetNcbieaa(), CSeq_data_Base::GetNcbistdaa(), CSeq_inst_Base::GetSeq_data(), i, int, CSeq_data_Base::IsIupacaa(), CSeq_data_Base::IsNcbieaa(), CSeq_data_Base::IsNcbistdaa(), CSeq_inst_Base::IsSetSeq_data(), and kEmptyStr.
Referenced by CDUpdater::update().
Definition at line 389 of file cuSequence.cpp.
References GetNcbieaaString(), and str().
char GetResidueAtPosition | ( | const CRef< CSeq_entry > & | seqEntry, |
int | pos, | ||
bool | zeroBased | ||
) |
Definition at line 380 of file cuSequence.cpp.
References CBioseq_Base::GetInst(), CSeq_entry_Base::GetSeq(), CSeq_entry_Base::IsSeq(), and CSeq_inst_Base::IsSetSeq_data().
Definition at line 216 of file cuSequence.cpp.
References CAliasBase< TPrim >::Get(), CBioseq_Base::GetInst(), CSeq_data_Base::GetIupacaa(), CSeq_inst_Base::GetLength(), CSeq_data_Base::GetNcbieaa(), CSeq_data_Base::GetNcbistdaa(), CSeq_inst_Base::GetSeq_data(), CSeq_data_Base::IsIupacaa(), CSeq_data_Base::IsNcbieaa(), CSeq_data_Base::IsNcbistdaa(), CSeq_inst_Base::IsSetLength(), CSeq_inst_Base::IsSetSeq_data(), and len.
Referenced by GetSeqLength(), CDUpdater::isFragmentedSeq(), CSimpleB2SWrapper::SetSeq(), CWGSSeqIterator::x_AddQualityChunkInfo(), and CWGSSeqIterator::x_InitSplit().
bool GetSeqLength | ( | const CRef< CSeq_entry > & | Seq, |
int & | len | ||
) |
Definition at line 242 of file cuSequence.cpp.
References CRef< C, Locker >::Empty(), CBioseq_Base::GetInst(), CSeq_inst_Base::GetLength(), CSeq_entry_Base::GetSeq(), GetSeqLength(), CSeq_entry_Base::IsSet(), CSeq_inst_Base::IsSetLength(), and len.
Definition at line 190 of file cuSequence.cpp.
References CSeq_descr_Base::Get(), CBioseq_Base::GetDescr(), CBioseq_Base::IsSetDescr(), and kEmptyStr.
Referenced by CCdCore::GetSpeciesForIndex(), and CCdCore::GetSpeciesForRow().
Definition at line 139 of file cuSequence.cpp.
References CSeq_descr_Base::Get(), COrg_ref_Base::GetDb(), CBioseq_Base::GetDescr(), CBioseq_Base::IsSetDescr(), NULL, TAX_ID_FROM, and ZERO_TAX_ID.
Referenced by GetCommonTax(), and IsEnvironmentalSeq().
bool HasSeqIdOfType | ( | const CBioseq & | bioseq, |
CSeq_id::E_Choice | choice | ||
) |
Definition at line 559 of file cuSequence.cpp.
References CBioseq_Base::GetId(), and result.
Referenced by HasSeqIdOfType().
bool HasSeqIdOfType | ( | const CRef< CSeq_entry > & | seqEntry, |
CSeq_id::E_Choice | choice | ||
) |
Definition at line 572 of file cuSequence.cpp.
References CSeq_entry_Base::GetSeq(), CBioseq_set_Base::GetSeq_set(), CSeq_entry_Base::GetSet(), HasSeqIdOfType(), CSeq_entry_Base::IsSeq(), CSeq_entry_Base::IsSet(), CRef< C, Locker >::NotEmpty(), and result.
Definition at line 405 of file cuSequence.cpp.
References CSeq_id_Base::GetLocal(), CObject_id_Base::GetStr(), CSeq_id_Base::IsLocal(), CObject_id_Base::IsStr(), CRef< C, Locker >::NotEmpty(), and result.
Referenced by ConsensusMaker::addRows(), ConsensusMaker::ConsensusMaker(), CCdCore::FindConsensusInSequenceList(), CCdCore::GetRowsWithConsensus(), CCdCore::HasConsensusSequence(), PssmMaker::printAlignment(), PssmMaker::printAlignmentByColumn(), PssmMaker::PssmMaker(), and CCdCore::UsesConsensusSequenceAsMaster().
Definition at line 184 of file cuSequence.cpp.
References ENVIRONMENTAL_SEQUENCE_TAX_ID, and GetTaxIdInBioseq().
bool NcbieaaToNcbistdaaString | ( | const std::string & | str, |
vector< char > & | vec | ||
) |
Definition at line 272 of file cuSequence.cpp.
References CSeqConvert::Convert(), CSeqUtil::e_Ncbieaa, CSeqUtil::e_Ncbistdaa, result, and str().
void NcbistdaaToNcbieaaString | ( | const std::vector< char > & | vec, |
std::string * | str | ||
) |
Definition at line 258 of file cuSequence.cpp.
References CSeqConvert::Convert(), CSeqUtil::e_Ncbieaa, CSeqUtil::e_Ncbistdaa, and str().
Referenced by GetNcbieaaString().
Definition at line 80 of file cuSequence.cpp.
References Empty(), CBioseq_Base::GetId(), result, and SeqIdsMatch().
Referenced by CCdCore::GetSeqIndex().
Definition at line 70 of file cuSequence.cpp.
Referenced by CheckSeqIdInDD(), CDUpdater::findSeq(), GetAlignedPositions(), CCdCore::GetNthMatchFor(), AlignmentCollection::GetRowsWithSeqID(), CCdCore::HasSeqId(), BlockModel::isAlike(), BlockModel::mask(), BlockModel::overlap(), CDUpdater::overlapWithCDRow(), BlockModelPair::remaster(), and SeqIdHasMatchInBioseq().
void SimplifyBioseqForCD | ( | CBioseq & | bioseq, |
const vector< string > & | keptComments, | ||
bool | keepPDBBlock | ||
) |
Definition at line 718 of file cuSequence.cpp.
References CPDB_block_Base::CanGetCompound(), CSeq_descr_Base::Get(), CPDB_block_Base::GetCompound(), CBioseq_Base::GetDescr(), CSeq_descr_Base::IsSet(), kEmptyStr, CBioseq_Base::ResetAnnot(), CBioseq_Base::ResetDescr(), CSeq_descr_Base::Set(), CBioseq_Base::SetDescr(), and CSeqdesc_Base::SetTitle().
Referenced by SimplifySeqEntryForCD().
void SimplifySeqEntryForCD | ( | CRef< CSeq_entry > & | seqEntry, |
const vector< string > & | keptComments, | ||
bool | keepPDBBlock | ||
) |
Definition at line 778 of file cuSequence.cpp.
References CRef< C, Locker >::Empty(), CSeq_entry_Base::IsSeq(), CSeq_entry_Base::IsSet(), CSeq_entry_Base::SetSeq(), CBioseq_set_Base::SetSeq_set(), CSeq_entry_Base::SetSet(), and SimplifyBioseqForCD().