#include <ncbi_pch.hpp>
#include <corelib/ncbiapp.hpp>
#include <corelib/ncbienv.hpp>
#include <corelib/ncbiargs.hpp>
#include <algo/gnomon/chainer.hpp>
#include <algo/gnomon/id_handler.hpp>
#include <algo/gnomon/gnomon_exception.hpp>
#include <algo/gnomon/glb_align.hpp>
#include <util/sequtil/sequtil_manip.hpp>
#include <algo/gnomon/gnomon_model.hpp>
#include <algo/gnomon/gnomon.hpp>
#include <algo/gnomon/annot.hpp>
#include <map>
#include <sstream>
#include <tuple>
#include <unordered_set>
#include <unordered_map>
#include <objects/general/Object_id.hpp>
#include <objmgr/object_manager.hpp>
#include <objmgr/feat_ci.hpp>
#include <objmgr/util/sequence.hpp>
#include "gnomon_seq.hpp"

Include dependency graph for chainer.cpp:

Go to the SVN repository for this file.

Classes
class	CChainer::CChainerImpl

struct	SChainMember

class	CChain

class	CGene

struct	GenomeOrderD

struct	AlignIdOrder

struct	LeftOrder

struct	LeftOrderD

struct	RightOrder

struct	RightOrderD

struct	CdsNumOrder

struct	ScoreOrder

class	CChainMembers

struct	GModelOrder

struct	AlignSeqOrder

struct	AlignLenOrder

struct	SLinker

struct	RangeOrder

struct	s_ByAccVerLen

struct	OverlapsSameAccessionAlignment

struct	ConnectsParalogs

struct	SFShiftsCluster

struct	ProjectCDS

struct	TrimAlignment

struct	DoNotBelieveShortPolyATail

struct	DoNotBelieveFrameShiftsWithoutCdsEvidence

Macros
#define	START_BONUS 600

#define	NON_CDNA_INTRON_PENALTY 20

#define	PGAP_PENALTY 120

#define	SCAN_WINDOW 49

#define	MIN_UTR_EXON 15

#define	COVERAGE_DROP 0.1

#define	COVERAGE_BUMP 3

#define	SMALL_GAP_UTR 100

#define	PROT_CLIP 120

#define	PROT_CLIP_FRAC 0.20

#define	MIN_PART 30

#define	BLOCK_OF_Ns 35

Typedefs
typedef list< CChain >	TChainList

typedef list< CChain * >	TChainPointerList

typedef vector< SChainMember * >	TContained

typedef map< Int8, CAlignModel * >	TOrigAligns

typedef map< Int8, CGeneModel >	TUnmodAligns

typedef set< SChainMember * >	TMemberPtrSet

typedef CChain *	TChainPtr

typedef vector< pair< SChainMember , CGene > >	TMemeberGeneVec

typedef tuple< Int8, TSignedSeqRange >	TIdLim

typedef vector< SLinker >	TLinkers

typedef set< TSignedSeqRange, RangeOrder >	TRangePrecedeSet

Enumerations
enum	{ eCDS , eLeftUTR , eRightUTR }

Functions
bool	BelongToExon (const CGeneModel::TExons &exons, int pos)

static bool	DescendingModelOrder (const CChain &a, const CChain &b)

static bool	DescendingModelOrderP (const TChainPtr &a, const TChainPtr &b)

static bool	DescendingModelOrderPConsistentCoverage (const TChainPtr &a, const TChainPtr &b)

TIdLim	AlignIdLimits (SChainMember *mp)

template<class C >
void	uniq (C &container)

TSignedSeqRange	ExtendedMaxCdsLimits (const CGeneModel &a, const CCDSInfo &cds)

TInDels	StrictlyContainedInDels (const TInDels &indels, const TSignedSeqRange &lim)

bool	MemberIsCoding (const SChainMember *mp)

bool	MemberIsMarkedForDeletion (const SChainMember *mp)

string	GetLinkedIdsForMember (const SChainMember &mi)

bool	GoodSupportForIntrons (const CGeneModel &chain, const SMinScor &minscor, map< TSignedSeqRange, int > &mrna_count, map< TSignedSeqRange, int > &est_count, map< TSignedSeqRange, int > &rnaseq_count)

void	MarkUnwantedLowSupportIntrons (TContained &pointers, const SMinScor &minscor, map< TSignedSeqRange, int > &mrna_count, map< TSignedSeqRange, int > &est_count, map< TSignedSeqRange, int > &rnaseq_count)

pair< string, int >	GetAccVer (const CAlignModel &a, CScope &scope)

static int	s_ExonLen (const CGeneModel &a)

string	FindMultiplyIncluded (CAlignModel &algn, TAlignModelList &clust)

double	InframeFraction (const CGeneModel &a, TSignedSeqPos left, TSignedSeqPos right)

bool	LeftAndLongFirst (const CGeneModel &a, const CGeneModel &b)

bool	OverlappingIndel (int pos, const CInDelInfo &indl)

TInDels	CombineCorrectionsAndIndels (const TSignedSeqRange exona, int extra_left, int extra_right, const TSignedSeqRange exonb, const TInDels &editing_indels_frombtoa, const TInDels &exona_indels)

int	EffectiveExonLength (const CModelExon &e, const CAlignMap &alignmap, bool snap_to_codons)

Macro Definition Documentation

◆ BLOCK_OF_Ns

#define BLOCK_OF_Ns 35

◆ COVERAGE_BUMP

#define COVERAGE_BUMP 3

Definition at line 5745 of file chainer.cpp.

◆ COVERAGE_DROP

#define COVERAGE_DROP 0.1

Definition at line 5744 of file chainer.cpp.

◆ MIN_PART

#define MIN_PART 30

Definition at line 6618 of file chainer.cpp.

◆ MIN_UTR_EXON

#define MIN_UTR_EXON 15

Definition at line 5470 of file chainer.cpp.

◆ NON_CDNA_INTRON_PENALTY

#define NON_CDNA_INTRON_PENALTY 20

Definition at line 2568 of file chainer.cpp.

◆ PGAP_PENALTY

#define PGAP_PENALTY 120

◆ PROT_CLIP

#define PROT_CLIP 120

Definition at line 6616 of file chainer.cpp.

◆ PROT_CLIP_FRAC

#define PROT_CLIP_FRAC 0.20

Definition at line 6617 of file chainer.cpp.

◆ SCAN_WINDOW

#define SCAN_WINDOW 49

Definition at line 4365 of file chainer.cpp.

◆ SMALL_GAP_UTR

#define SMALL_GAP_UTR 100

Definition at line 5746 of file chainer.cpp.

◆ START_BONUS

#define START_BONUS 600

Definition at line 1725 of file chainer.cpp.

Typedef Documentation

◆ TChainList

typedef list<CChain> TChainList

Definition at line 73 of file chainer.cpp.

◆ TChainPointerList

typedef list<CChain*> TChainPointerList

Definition at line 74 of file chainer.cpp.

◆ TChainPtr

typedef CChain* TChainPtr

Definition at line 699 of file chainer.cpp.

◆ TContained

typedef vector<SChainMember*> TContained

Definition at line 78 of file chainer.cpp.

◆ TIdLim

typedef tuple<Int8, TSignedSeqRange> TIdLim

Definition at line 1257 of file chainer.cpp.

◆ TLinkers

typedef vector<SLinker> TLinkers

Definition at line 4679 of file chainer.cpp.

◆ TMemberPtrSet

typedef set<SChainMember*> TMemberPtrSet

Definition at line 226 of file chainer.cpp.

◆ TMemeberGeneVec

typedef vector< pair<SChainMember*,CGene*> > TMemeberGeneVec

Definition at line 1255 of file chainer.cpp.

◆ TOrigAligns

typedef map<Int8,CAlignModel*> TOrigAligns

Definition at line 80 of file chainer.cpp.

◆ TRangePrecedeSet

typedef set<TSignedSeqRange,RangeOrder> TRangePrecedeSet

Definition at line 4686 of file chainer.cpp.

◆ TUnmodAligns

typedef map<Int8,CGeneModel> TUnmodAligns

Definition at line 81 of file chainer.cpp.

Enumeration Type Documentation

◆ anonymous enum

anonymous enum

Enumerator
eCDS
eLeftUTR
eRightUTR

Definition at line 220 of file chainer.cpp.

Function Documentation

◆ AlignIdLimits()

TIdLim AlignIdLimits ( SChainMember * mp )

Definition at line 1258 of file chainer.cpp.

References CGeneModel::ID(), CGeneModel::Limits(), and SChainMember::m_align.

Referenced by AlignIdOrder::operator()(), and CChainer::CChainerImpl::TrimAlignmentsIncludedInDifferentGenes().

◆ BelongToExon()

bool BelongToExon	(	const CGeneModel::TExons &	exons,
		int	pos
	)

Definition at line 64 of file chainer.cpp.

References i, Include(), and ITERATE.

◆ CombineCorrectionsAndIndels()

TInDels CombineCorrectionsAndIndels	(	const TSignedSeqRange	exona,
		int	extra_left,
		int	extra_right,
		const TSignedSeqRange	exonb,
		const TInDels &	editing_indels_frombtoa,
		const TInDels &	exona_indels
	)

Definition at line 7689 of file chainer.cpp.

References _ASSERT, count, CInDelInfo::eDel, CInDelInfo::eGenomeNotCorrect, CInDelInfo::eIns, ERASE_ITERATE, CRange_Base::GetFrom(), COpenRange< Position >::GetLength(), CRange_Base::GetTo(), ip, len, max(), min(), OverlappingIndel(), and pb.

Referenced by CGnomonAnnotator_Base::MapOneModelToEditedContig(), and CGnomonAnnotator_Base::MapOneModelToOrigContig().

◆ DescendingModelOrder()

static bool DescendingModelOrder	(	const CChain &	a,
		const CChain &	b
	)

static

Definition at line 611 of file chainer.cpp.

References a, b, CGeneModel::eCap, CGeneModel::ecDNAIntrons, CGeneModel::ePolyA, and fabs.

Referenced by DescendingModelOrderP(), DescendingModelOrderPConsistentCoverage(), and CChainer::CChainerImpl::FindAltsForGeneSeeds().

◆ DescendingModelOrderP()

static bool DescendingModelOrderP	(	const TChainPtr &	a,
		const TChainPtr &	b
	)

static

Definition at line 700 of file chainer.cpp.

References a, b, and DescendingModelOrder().

Referenced by CChainer::CChainerImpl::FilterOutSimilarsWithLowerScore(), CChainer::CChainerImpl::FindGeneSeeds(), CChainer::CChainerImpl::PlaceAllYouCan(), and CChainer::CChainerImpl::ReplacePseudoGeneSeeds().

◆ DescendingModelOrderPConsistentCoverage()

static bool DescendingModelOrderPConsistentCoverage	(	const TChainPtr &	a,
		const TChainPtr &	b
	)

static

Definition at line 704 of file chainer.cpp.

References a, b, DescendingModelOrder(), and CGeneModel::eConsistentCoverage.

Referenced by CChainer::CChainerImpl::FindAltsForGeneSeeds().

◆ EffectiveExonLength()

int EffectiveExonLength	(	const CModelExon &	e,
		const CAlignMap &	alignmap,
		bool	snap_to_codons
	)

Definition at line 8617 of file chainer.cpp.

References CAlignMap::FShiftedLen(), COpenRange< Position >::GetLength(), min(), and CAlignMap::ShrinkToRealPoints().

Referenced by CutShortPartialExons::transform_align().

◆ ExtendedMaxCdsLimits()

TSignedSeqRange ExtendedMaxCdsLimits	(	const CGeneModel &	a,
		const CCDSInfo &	cds
	)

Definition at line 2055 of file chainer.cpp.

References a, limits, and CCDSInfo::MaxCdsLimits().

Referenced by CChainer::CChainerImpl::CanIncludeJinI(), CChainer::CChainerImpl::LRCanChainItoJ(), and CChainer::CChainerImpl::RightLeft().

◆ FindMultiplyIncluded()

string FindMultiplyIncluded	(	CAlignModel &	algn,
		TAlignModelList &	clust
	)

Definition at line 6534 of file chainer.cpp.

References CGeneModel::AlignLen(), CGeneModel::Continuous(), CGeneModel::eProt, CGeneModel::IdenticalAlign(), set< Key, Compare >::insert(), CGeneModel::isCompatible(), kEmptyStr, len, NON_CONST_ITERATE, CGeneModel::Strand(), CAlignModel::TargetAccession(), and CGeneModel::Type().

Referenced by ConnectsParalogs::align_predicate().

◆ GetAccVer()

pair<string,int> GetAccVer	(	const CAlignModel &	a,
		CScope &	scope
	)

Definition at line 6323 of file chainer.cpp.

References a, CSeq_id_Handle::AsString(), eGetId_ForceAcc, CGeneModel::eProt, CTextseq_id_Base::GetAccession(), GetId(), CSeq_id_Handle::GetSeqId(), CSeq_id::GetTextseq_Id(), CTextseq_id_Base::GetVersion(), CTextseq_id_Base::IsSetAccession(), and CTextseq_id_Base::IsSetVersion().

Referenced by s_ByAccVerLen::operator()(), and OverlapsSameAccessionAlignment::OverlapsSameAccessionAlignment().

◆ GetLinkedIdsForMember()

string GetLinkedIdsForMember ( const SChainMember & mi )

Definition at line 2984 of file chainer.cpp.

References CGeneModel::ID(), ITERATE, SChainMember::m_align, SChainMember::m_left_member, SChainMember::m_right_member, and ct::sort().

Referenced by CChainer::CChainerImpl::CreateChainsForPartialProteins().

◆ GoodSupportForIntrons()

bool GoodSupportForIntrons	(	const CGeneModel &	chain,
		const SMinScor &	minscor,
		map< TSignedSeqRange, int > &	mrna_count,
		map< TSignedSeqRange, int > &	est_count,
		map< TSignedSeqRange, int > &	rnaseq_count
	)

Definition at line 3001 of file chainer.cpp.

References CGeneModel::Exons(), i, int, SMinScor::m_minsupport, SMinScor::m_minsupport_mrna, and SMinScor::m_minsupport_rnaseq.

Referenced by MarkUnwantedLowSupportIntrons().

◆ InframeFraction()

double InframeFraction	(	const CGeneModel &	a,
		TSignedSeqPos	left,
		TSignedSeqPos	right
	)

Definition at line 6895 of file chainer.cpp.

References a, CAlignMap::FShiftedLen(), ITERATE, len, prev(), and CAlignMap::ShrinkToRealPoints().

Referenced by ProjectCDS::transform_align().

◆ LeftAndLongFirst()

bool LeftAndLongFirst	(	const CGeneModel &	a,
		const CGeneModel &	b
	)

Definition at line 7274 of file chainer.cpp.

References a, and b.

◆ MarkUnwantedLowSupportIntrons()

void MarkUnwantedLowSupportIntrons	(	TContained &	pointers,
		const SMinScor &	minscor,
		map< TSignedSeqRange, int > &	mrna_count,
		map< TSignedSeqRange, int > &	est_count,
		map< TSignedSeqRange, int > &	rnaseq_count
	)

Definition at line 3015 of file chainer.cpp.

References GoodSupportForIntrons(), i, and NON_CONST_ITERATE.

◆ MemberIsCoding()

bool MemberIsCoding ( const SChainMember * mp )

Definition at line 2975 of file chainer.cpp.

References BadScore(), SChainMember::m_cds_info, and CCDSInfo::Score().

Referenced by CChainer::CChainerImpl::MakeChains().

◆ MemberIsMarkedForDeletion()

bool MemberIsMarkedForDeletion ( const SChainMember * mp )

Definition at line 2979 of file chainer.cpp.

References SChainMember::m_marked_for_deletion.

Referenced by CChainer::CChainerImpl::MakeChains().

◆ OverlappingIndel()

bool OverlappingIndel	(	int	pos,
		const CInDelInfo &	indl
	)

Definition at line 7680 of file chainer.cpp.

References CInDelInfo::InDelEnd(), and CInDelInfo::IsDeletion().

Referenced by CombineCorrectionsAndIndels(), CGnomonAnnotator_Base::MapModelsToOrigContig(), CGnomonAnnotator_Base::MapOneModelToEditedContig(), and CGnomonAnnotator_Base::MapOneModelToOrigContig().

◆ s_ExonLen()

static int s_ExonLen ( const CGeneModel & a )

static

Definition at line 6374 of file chainer.cpp.

References a, ITERATE, and len.

Referenced by s_ByAccVerLen::operator()().

◆ StrictlyContainedInDels()

TInDels StrictlyContainedInDels	(	const TInDels &	indels,
		const TSignedSeqRange &	lim
	)

Definition at line 2358 of file chainer.cpp.

References CRange_Base::GetFrom(), CRange_Base::GetTo(), and ITERATE.

Referenced by CChainer::CChainerImpl::CanIncludeJinI(), CChainer::CChainerImpl::CombineCompatibleChains(), CChainer::CChainerImpl::FindOptimalChainForProtein(), CChainer::FindSelenoproteinsClipProteinsToStartStop(), CChainer::CChainerImpl::LRCanChainItoJ(), and CChainer::CChainerImpl::RightLeft().

◆ uniq()

template<class C >

void uniq ( C & container )

Definition at line 1932 of file chainer.cpp.

References ct::sort().

Referenced by CChainer::CChainerImpl::AddIfCompatible(), CChainer::CChainerImpl::MakeChains(), CChainer::CChainerImpl::ReplicatePStops(), DeBruijn::CKmerCountTemplate< LargeInt< 1 >, LargeInt< 2 >, LargeInt< 4 >, LargeInt< 8 >, LargeInt< 16 > >::SortAndExtractUniq(), DeBruijn::CKmerCountTemplate< LargeInt< 1 >, LargeInt< 2 >, LargeInt< 4 >, LargeInt< 8 >, LargeInt< 16 > >::SortAndUniq(), and CDiscrepancyVisitorImpl< _Name >::Summarize().

Classes

Macros

Typedefs

Enumerations

Functions

Macro Definition Documentation

◆ BLOCK_OF_Ns

◆ COVERAGE_BUMP

◆ COVERAGE_DROP

◆ MIN_PART

◆ MIN_UTR_EXON

◆ NON_CDNA_INTRON_PENALTY

◆ PGAP_PENALTY

◆ PROT_CLIP

◆ PROT_CLIP_FRAC

◆ SCAN_WINDOW

◆ SMALL_GAP_UTR

◆ START_BONUS

Typedef Documentation

◆ TChainList

◆ TChainPointerList

◆ TChainPtr

◆ TContained

◆ TIdLim

◆ TLinkers

◆ TMemberPtrSet

◆ TMemeberGeneVec

◆ TOrigAligns

◆ TRangePrecedeSet

◆ TUnmodAligns

Enumeration Type Documentation

◆ anonymous enum

Function Documentation

◆ AlignIdLimits()

◆ BelongToExon()

◆ CombineCorrectionsAndIndels()

◆ DescendingModelOrder()

◆ DescendingModelOrderP()

◆ DescendingModelOrderPConsistentCoverage()

◆ EffectiveExonLength()

◆ ExtendedMaxCdsLimits()

◆ FindMultiplyIncluded()

◆ GetAccVer()

◆ GetLinkedIdsForMember()

◆ GoodSupportForIntrons()

◆ InframeFraction()

◆ LeftAndLongFirst()

◆ MarkUnwantedLowSupportIntrons()

◆ MemberIsCoding()

◆ MemberIsMarkedForDeletion()

◆ OverlappingIndel()

◆ s_ExonLen()

◆ StrictlyContainedInDels()

◆ uniq()