score_builder_base.cpp File Reference

#include <ncbi_pch.hpp>
#include <objtools/alnmgr/score_builder_base.hpp>
#include <objtools/alnmgr/alntext.hpp>
#include <util/sequtil/sequtil_manip.hpp>
#include <objtools/alnmgr/alnvec.hpp>
#include <objtools/alnmgr/pairwise_aln.hpp>
#include <objtools/alnmgr/aln_converters.hpp>
#include <objmgr/objmgr_exception.hpp>
#include <objmgr/seq_vector.hpp>
#include <objmgr/feat_ci.hpp>
#include <objects/seqloc/Seq_loc.hpp>
#include <objects/seq/Annot_descr.hpp>
#include <objects/seq/Annotdesc.hpp>
#include <objects/seqalign/Seq_align.hpp>
#include <objects/seqalign/Std_seg.hpp>
#include <objects/seqalign/Spliced_seg.hpp>
#include <objects/seqalign/Spliced_exon.hpp>
#include <objects/seqalign/Spliced_exon_chunk.hpp>
#include <objects/seqalign/Product_pos.hpp>
#include <objects/seqalign/Prot_pos.hpp>
#include <objmgr/util/sequence.hpp>
#include <objects/seqfeat/Org_ref.hpp>
#include <objects/seqfeat/OrgName.hpp>
#include <objects/seqfeat/Genetic_code_table.hpp>

Include dependency graph for score_builder_base.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Functions
	USING_SCOPE (objects)
static TSeqPos	s_IntersectionLength (const CRangeCollection< TSeqPos > &ranges, const TSeqRange &range)
	Get length of intersection between a range and a range collection. More...
static void	s_GetNucIdentityMismatch (const vector< string > &data, int *identities, int *mismatches)
	calculate mismatches and identities in a seq-align More...
static void	s_GetSplicedSegIdentityMismatch (CScope &scope, const CSeq_align &align, const CRangeCollection< TSeqPos > &ranges, int *identities, int *mismatches)
static void	s_GetCountIdentityMismatch (CScope &scope, const CSeq_align &align, int *identities, int *mismatches, const CRangeCollection< TSeqPos > &ranges=CRangeCollection< TSeqPos >(TSeqRange::GetWhole()))
static void	s_GetPercentIdentity (CScope &scope, const CSeq_align &align, int *identities, int *mismatches, double *pct_identity, CScoreBuilderBase::EPercentIdentityType type, const CRangeCollection< TSeqPos > &ranges=CRangeCollection< TSeqPos >(TSeqRange::GetWhole()))
	calculate the percent identity we also return the count of identities and mismatches More...
static bool	s_SequenceIsProtein (CScope &scope, const CSeq_id &id)
	calculate the percent coverage More...
static bool	s_IsProteinToGenomic (CScope &scope, const CSeq_align &align)
static void	s_GetPercentCoverage (CScope &scope, const CSeq_align &align, const CRangeCollection< TSeqPos > &ranges, double *pct_coverage, unsigned query=0)
string	GetDonor (const objects::CSpliced_exon &exon)
string	GetAcceptor (const objects::CSpliced_exon &exon)
bool	IsConsSplice (const string &donor, const string acc)

Function Documentation

GetAcceptor()

string GetAcceptor

(

const objects::CSpliced_exon &

exon

)

Definition at line 982 of file score_builder_base.cpp.

References string.

Referenced by CScoreBuilderBase::AddSplignScores().

GetDonor()

string GetDonor

(

const objects::CSpliced_exon &

exon

)

Definition at line 975 of file score_builder_base.cpp.

References string.

Referenced by CScoreBuilderBase::AddSplignScores(), and CSplign::x_Run().

IsConsSplice()

bool IsConsSplice	(	const string &	donor,
		const string	acc
	)

Definition at line 990 of file score_builder_base.cpp.

References toupper().

Referenced by CScoreBuilderBase::AddSplignScores().

s_GetCountIdentityMismatch()

static void s_GetCountIdentityMismatch ( CScope &  scope, const CSeq_align &  align, int *  identities, int *  mismatches, const CRangeCollection< TSeqPos > &  ranges = CRangeCollection<TSeqPos>(TSeqRange::GetWhole())  )

static

Definition at line 285 of file score_builder_base.cpp.

References _ASSERT, CSeq_id::AsFastaString(), CProduct_pos::AsSeqPos(), data, CSeq_align_Base::C_Segs::e_Denseg, CSpliced_exon_chunk_Base::e_Diag, CSeq_align_Base::C_Segs::e_Disc, CSpliced_exon_chunk_Base::e_Match, CSpliced_exon_chunk_Base::e_Mismatch, CSpliced_exon_chunk_Base::e_Product_ins, CSeq_align_Base::C_Segs::e_Spliced, CSeq_align_Base::C_Segs::e_Std, CSeq_align::eScore_IdentityCount, CSeq_align_set_Base::Get(), CSeq_align::GetAlignLength(), CSeq_align_Base::C_Segs::GetDenseg(), CSpliced_exon_chunk_Base::GetDiag(), CSeq_align_Base::C_Segs::GetDisc(), CSpliced_seg_Base::GetExons(), CRange_Base::GetFrom(), CSpliced_exon_chunk_Base::GetMatch(), CSpliced_exon_chunk_Base::GetMismatch(), CSeq_align::GetNamedScore(), CAlnMap::GetNumRows(), CAlnMap::GetNumSegs(), CSpliced_exon_Base::GetParts(), CSpliced_exon_Base::GetProduct_end(), CSpliced_exon_chunk_Base::GetProduct_ins(), CSpliced_exon_Base::GetProduct_start(), CSeq_align_Base::GetSegs(), CSeq_align::GetSeq_id(), CAlnVec::GetSeqString(), CSeq_align_Base::C_Segs::GetSpliced(), CAlnMap::GetStart(), CAlnMap::GetStop(), i, CRangeCollection< Position >::IntersectWith(), CSpliced_exon_Base::IsSetParts(), ITERATE, len, NCBI_RETHROW_SAME, NCBI_THROW, offset, s_GetNucIdentityMismatch(), s_GetSplicedSegIdentityMismatch(), s_IntersectionLength(), COpenRange< Position >::Set(), CSeq_align_Base::C_Segs::Which(), and CSpliced_exon_chunk_Base::Which().

Referenced by CScoreBuilderBase::GetIdentityCount(), CScoreBuilderBase::GetMismatchCount(), and s_GetPercentIdentity().

s_GetNucIdentityMismatch()

static void s_GetNucIdentityMismatch ( const vector< string > &  data, int *  identities, int *  mismatches  )

static

calculate mismatches and identities in a seq-align

Definition at line 93 of file score_builder_base.cpp.

References a, b, data, i, NCBI_THROW, and ncbi::grid::netcache::search::fields::size.

Referenced by s_GetCountIdentityMismatch().

s_GetPercentCoverage()

static void s_GetPercentCoverage ( CScope &  scope, const CSeq_align &  align, const CRangeCollection< TSeqPos > &  ranges, double *  pct_coverage, unsigned  query = 0  )

static

Definition at line 537 of file score_builder_base.cpp.

References CRangeCollection< Position >::begin(), CRangeCollection< Position >::empty(), eNa_strand_minus, CSeq_align::eScore_PercentCoverage, CSeq_align::GetAlignLengthWithinRanges(), CScope::GetBioseqHandle(), CRangeCollection< Position >::GetCoveredLength(), CSeq_align::GetNamedScore(), CSpliced_seg_Base::GetPoly_a(), CSpliced_seg_Base::GetProduct_length(), CSpliced_seg_Base::GetProduct_strand(), CSeq_align_Base::GetSegs(), CSeq_align::GetSeq_id(), CSeq_align_Base::C_Segs::GetSpliced(), CSpliced_seg_Base::IsSetPoly_a(), CSpliced_seg_Base::IsSetProduct_length(), CSpliced_seg_Base::IsSetProduct_strand(), CSeq_align_Base::C_Segs::IsSpliced(), CSeq_align_Base::C_Segs::IsStd(), NCBI_THROW, query, and s_IsProteinToGenomic().

Referenced by CScoreBuilderBase::ComputeScore(), and CScoreBuilderBase::GetPercentCoverage().

s_GetPercentIdentity()

static void s_GetPercentIdentity ( CScope &  scope, const CSeq_align &  align, int *  identities, int *  mismatches, double *  pct_identity, CScoreBuilderBase::EPercentIdentityType  type, const CRangeCollection< TSeqPos > &  ranges = CRangeCollection<TSeqPos>(TSeqRange::GetWhole())  )

static

calculate the percent identity we also return the count of identities and mismatches

Definition at line 453 of file score_builder_base.cpp.

References CScoreBuilderBase::eGapped, CScoreBuilderBase::eGBDNA, CScoreBuilderBase::eUngapped, CSeq_align::GetAlignLengthWithinRanges(), CSeq_align::GetNumGapOpeningsWithinRanges(), and s_GetCountIdentityMismatch().

Referenced by CScoreBuilderBase::AddScore(), CScoreBuilderBase::ComputeScore(), and CScoreBuilderBase::GetPercentIdentity().

s_GetSplicedSegIdentityMismatch()

static void s_GetSplicedSegIdentityMismatch ( CScope &  scope, const CSeq_align &  align, const CRangeCollection< TSeqPos > &  ranges, int *  identities, int *  mismatches  )

static

Definition at line 127 of file score_builder_base.cpp.

References CSeq_id::AsFastaString(), ConvertSeqAlignToPairwiseAln(), CSeqUtil::e_Iupacna, CBioseq_Handle::eCoding_Iupac, CSpliced_seg_Base::eProduct_type_protein, CSpliced_seg_Base::eProduct_type_transcript, gen, CScope::GetBioseqHandle(), CTrans_table::GetCodonResidue(), CRange_Base::GetFrom(), COrgName_Base::GetGcode(), GetOrg_ref(), COrg_ref_Base::GetOrgname(), CSpliced_seg_Base::GetProduct_type(), CSeq_align_Base::GetSegs(), CSeq_align::GetSeq_id(), CSeqVector::GetSeqData(), CSeq_align_Base::C_Segs::GetSpliced(), CTrans_table::GetStartResidue(), CRange_Base::GetTo(), CGen_code_table::GetTransTable(), i, CRangeCollection< Position >::IntersectWith(), ITERATE, match(), NCBI_THROW, compile_time_bits::range(), CSeqManip::ReverseComplement(), and CTrans_table::SetCodonState().

Referenced by s_GetCountIdentityMismatch().

s_IntersectionLength()

static TSeqPos s_IntersectionLength ( const CRangeCollection< TSeqPos > &  ranges, const TSeqRange &  range  )

inlinestatic

Get length of intersection between a range and a range collection.

Definition at line 79 of file score_builder_base.cpp.

References ITERATE, and compile_time_bits::range().

Referenced by s_GetCountIdentityMismatch().

s_IsProteinToGenomic()

static bool s_IsProteinToGenomic ( CScope &  scope, const CSeq_align &  align  )

static

Definition at line 506 of file score_builder_base.cpp.

References CSpliced_seg_Base::eProduct_type_protein, CSeq_align_Base::C_Segs::GetDenseg(), CSpliced_seg_Base::GetProduct_type(), CSeq_align_Base::GetSegs(), CSeq_align::GetSeq_id(), CSeq_align_Base::C_Segs::GetSpliced(), CSeq_align_Base::C_Segs::IsDenseg(), CDense_seg::IsSetWidths(), CSeq_align_Base::C_Segs::IsSpliced(), and s_SequenceIsProtein().

Referenced by s_GetPercentCoverage().

s_SequenceIsProtein()

static bool s_SequenceIsProtein ( CScope &  scope, const CSeq_id &  id  )

static

calculate the percent coverage

Definition at line 489 of file score_builder_base.cpp.

References CSeq_inst_Base::eMol_aa, CSeq_inst_Base::eMol_not_set, eUnknown, CScope::GetBioseqHandle(), CScope::GetSequenceType(), CBioseq_Handle::IsAa(), and NCBI_THROW.

Referenced by s_IsProteinToGenomic().

USING_SCOPE()

USING_SCOPE

(

objects

)