Formatting of BLAST results in XML2 form, using the BLAST XML2 specification. More...

#include <ncbi_pch.hpp>
#include <objmgr/object_manager.hpp>
#include <objects/seqloc/Seq_interval.hpp>
#include <objmgr/util/sequence.hpp>
#include <objects/seqloc/Seq_id.hpp>
#include <objects/seqalign/Dense_diag.hpp>
#include <objects/seqalign/Dense_seg.hpp>
#include <objects/seqalign/Std_seg.hpp>
#include <algo/blast/format/blastxml2_format.hpp>
#include <algo/blast/format/blastfmtutil.hpp>
#include <objtools/align_format/showdefline.hpp>
#include <objtools/align_format/align_format_util.hpp>
#include <objtools/blast/seqdb_reader/seqdb.hpp>
#include <objects/blastxml2/blastxml2__.hpp>
#include <serial/objostrxml.hpp>
#include <serial/objostrjson.hpp>
#include <algo/blast/api/version.hpp>
#include <sstream>
#include <algorithm>

Include dependency graph for blastxml2_format.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Classes
class	CBlastOStreamXml

class	CBlastOStreamJson

Functions
	USING_SCOPE (objects)

	USING_SCOPE (blast)

	USING_SCOPE (align_format)

static int	s_GetTranslationFrame (bool plus_strand, int start, int end, int seq_length)
	Returns translation frame given the strand, alignment endpoints and total sequence length. More...

static void	s_SeqAlignSetToXMLHsps (list< CRef< blastxml2::CHsp > > &xhsp_list, const CSeq_align_set &alnset, CRef< CScope > scope, const CBlastFormattingMatrix *matrix, const ncbi::TMaskedQueryRegions &mask_info, int master_gentic_code, int slave_genetic_code)
	Creates a list of blastxml2::CHsp structures for the XML output, given a list of Seq-aligns. More...

static void	s_SeqAlignToXMLHit (CRef< blastxml2::CHit > &hit, const CSeq_align &align_in, CRef< CScope > scope, const CBlastFormattingMatrix *matrix, const ncbi::TMaskedQueryRegions &mask_info, bool ungapped, int master_gentice_code, int slave_genetic_code, bool hasTaxDB)
	Fill the blastxml2::CHit object in BLAST XML output, given an alignment and other information. More...

static const CSeq_id *	s_GetSubjectId (const CSeq_align &align)
	Retrieves subject Seq-id from a Seq-align. More...

static void	s_SetBlastXMlHitList (list< CRef< blastxml2::CHit > > &hits, const IBlastXML2ReportData *data, int num)
	Fills the list of blastxml2::CHit objects, given a list of Seq-aligns. More...

static void	s_SetBlastXMLParameters (blastxml2::CParameters &params, const IBlastXML2ReportData *data)
	Fills the parameters part of the BLAST XML output. More...

static void	s_SetBlastXMLStatistics (blastxml2::CStatistics &stats, const IBlastXML2ReportData *data, int num)
	Fills the search statistics part of the BLAST XML output for all queries. More...

static void	s_SetBlastXMLSearch (blastxml2::CSearch &search, const IBlastXML2ReportData *data, int num)

static CReference::EPublication	s_GetBlastPublication (EProgram program)
	Given BLAST task, returns enumerated value for the publication to be referenced. More...

static void	s_FillBlastOutput (blastxml2::CBlastOutput2 &bxmlout, const IBlastXML2ReportData *data)

static void	s_WriteXML2ObjectNoHeader (blastxml2::CBlastOutput2 &bxmlout, CNcbiOstream *out_stream)

static void	s_WriteXML2Object (blastxml2::CBlastOutput2 &bxmlout, CNcbiOstream *out_stream)

void	BlastXML2_FormatReport (const IBlastXML2ReportData data, CNcbiOstream out_stream)
	Fills all fields in the data structure for a BLAST XML report. More...

void	BlastXML2_FormatReport (const IBlastXML2ReportData *data, string file_name)

void	BlastXML2_PrintHeader (CNcbiOstream *out_stream)

void	BlastXML2_FormatError (int exit_code, string err_msg, CNcbiOstream *out_stream)

void	BlastJSON_PrintHeader (CNcbiOstream *out_stream)

static void	s_WriteJSONObjectNoHeader (blastxml2::CBlastOutput2 &bxmlout, CNcbiOstream *out_stream)

static void	s_WriteJSONObject (blastxml2::CBlastOutput2 &bxmlout, CNcbiOstream *out_stream)

void	BlastJSON_FormatReport (const IBlastXML2ReportData *data, string file_name)

void	BlastJSON_FormatReport (const IBlastXML2ReportData data, CNcbiOstream out_stream)

Detailed Description

Formatting of BLAST results in XML2 form, using the BLAST XML2 specification.

Definition in file blastxml2_format.cpp.

Function Documentation

◆ s_FillBlastOutput()

static void s_FillBlastOutput	(	blastxml2::CBlastOutput2 &	bxmlout,
		const IBlastXML2ReportData *	data
	)

static

Definition at line 656 of file blastxml2_format.cpp.

References CBlastFormatUtil::BlastGetVersion(), data, eUnknown, CReference::GetString(), i, NCBI_THROW, NULL, s_GetBlastPublication(), s_SetBlastXMLParameters(), and s_SetBlastXMLSearch().

Referenced by BlastJSON_FormatReport(), and BlastXML2_FormatReport().

◆ s_GetBlastPublication()

static CReference::EPublication s_GetBlastPublication ( EProgram program )

static

Given BLAST task, returns enumerated value for the publication to be referenced.

Parameters

program BLAST task [in]

Returns: What publication to reference?

Definition at line 637 of file blastxml2_format.cpp.

References CReference::eCompBasedStats, eDeltaBlast, CReference::eDeltaBlast, CReference::eGappedBlast, CReference::eMaxPublications, eMegablast, CReference::eMegaBlast, CReference::ePhiBlast, ePHIBlastn, ePHIBlastp, and ePSIBlast.

Referenced by s_FillBlastOutput().

◆ s_GetSubjectId()

static const CSeq_id* s_GetSubjectId ( const CSeq_align & align )

static

Retrieves subject Seq-id from a Seq-align.

Parameters

align Seq-align object [in]

Returns: Subject Seq-id for this Seq-align.

Definition at line 440 of file blastxml2_format.cpp.

References CSeq_align_Base::C_Segs::GetDendiag(), CSeq_align_Base::C_Segs::GetDenseg(), CDense_seg_Base::GetIds(), CSeq_align_Base::GetSegs(), CSeq_align_Base::C_Segs::GetStd(), CSeq_align_Base::C_Segs::IsDendiag(), CSeq_align_Base::C_Segs::IsDenseg(), CSeq_align_Base::C_Segs::IsStd(), and NULL.

Referenced by s_SetBlastXMlHitList().

◆ s_GetTranslationFrame()

static int s_GetTranslationFrame	(	bool	plus_strand,
		int	start,
		int	end,
		int	seq_length
	)

static

Returns translation frame given the strand, alignment endpoints and total sequence length.

Parameters

plus_strand	Is this position on a forward strand? [in]
start	Starting position, in 1-offset coordinates. [in]
end	Ending position in 1-offset coordinates [in]
seq_length	Total length of sequence [in]

Returns: Frame number.

Definition at line 74 of file blastxml2_format.cpp.

Referenced by s_SeqAlignSetToXMLHsps().

◆ s_SeqAlignSetToXMLHsps()

static void s_SeqAlignSetToXMLHsps	(	list< CRef< blastxml2::CHsp > > &	xhsp_list,
		const CSeq_align_set &	alnset,
		CRef< CScope >	scope,
		const CBlastFormattingMatrix *	matrix,
		const ncbi::TMaskedQueryRegions &	mask_info,
		int	master_gentic_code,
		int	slave_genetic_code
	)

static

Creates a list of blastxml2::CHsp structures for the XML output, given a list of Seq-aligns.

Parameters

xhsp_list	List of blastxml2::CHsp's to populate [in] [out]
alnset	Set of alignments to get data from [in]
scope	Scope for retrieving sequences [in]
matrix	256x256 matrix for calculating positives for a protein search. NULL is passed for a nucleotide search.
mask_info	Masking locations [in]

Definition at line 96 of file blastxml2_format.cpp.

References CDense_seg::Assign(), CreateDensegFromDendiag(), CSeq_align::CreateDensegFromStdseg(), CSeq_align::CreateTranslatedDensegFromNADenseg(), CDisplaySeqalign::eLowerCase, eNa_strand_minus, CSeq_align_set_Base::Get(), CSeq_align::GetAlignLength(), CScope::GetBioseqHandle(), CBioseq_Handle::GetBioseqLength(), CSeq_align_Base::C_Segs::GetDenseg(), CSeq_align_Base::GetSegs(), CSeq_align::GetSeq_id(), CSeq_align::GetSeqStart(), CSeq_align::GetSeqStop(), CSeq_align::GetSeqStrand(), CDense_seg_Base::GetStrands(), CSeq_align::GetTotalGapCount(), CBlastFormatUtil::GetWholeAlnSeqStrings(), i, CSeq_align_Base::C_Segs::IsDendiag(), CBioseq_Handle::IsNa(), CDense_seg_Base::IsSetStrands(), CSeq_align_Base::C_Segs::IsStd(), ITERATE, min(), NcbiEmptyString, CRef< C, Locker >::Reset(), CDense_seg::Reverse(), s_GetTranslationFrame(), and tmp.

Referenced by s_SeqAlignToXMLHit().

◆ s_SeqAlignToXMLHit()

static void s_SeqAlignToXMLHit	(	CRef< blastxml2::CHit > &	hit,
		const CSeq_align &	align_in,
		CRef< CScope >	scope,
		const CBlastFormattingMatrix *	matrix,
		const ncbi::TMaskedQueryRegions &	mask_info,
		bool	ungapped,
		int	master_gentice_code,
		int	slave_genetic_code,
		bool	hasTaxDB
	)

static

Fill the blastxml2::CHit object in BLAST XML output, given an alignment and other information.

Parameters

hit	blastxml2::CHit object to fill [in] [out]
align_in	Sequence alignment [in]
scope	Scope for retrieving sequences [in]
matrix	ASCII-alphabet matrix for calculation of positives [in]
mask_info	List of masking locations [in]
ungapped	Is this an ungapped search? [in]
master_genetic_code	query genetic code [in]
slave_genetic_code	subject genetic code [in]
hasTaxDB	Have access to taxonomy file [in]

Definition at line 347 of file blastxml2_format.cpp.

References _ASSERT, CSeq_id::AsFastaString(), CSeqDB::ExtractBlastDefline(), FindBestChoice(), CBlast_def_line_set_Base::Get(), CSeq_align_set_Base::Get(), CScope::GetBioseqHandle(), CBioseq_Handle::GetBioseqLength(), CSeq_align_Base::C_Segs::GetDisc(), CSeq_id_Handle::GetHandle(), CAlignFormatUtil::GetLabel(), GetLength(), CSeq_align_Base::GetSegs(), CBlast_def_line_Base::GetSeqid(), CSeq_id_Handle::GetSeqId(), CShowBlastDefline::GetSeqIdList(), CShowBlastDefline::GetSeqIdListString(), CBlast_def_line_Base::GetTaxid(), CSeqDB::GetTaxInfo(), CBlast_def_line_Base::GetTitle(), CSeq_align_Base::C_Segs::IsDisc(), CSeq_id_Base::IsLocal(), CBlast_def_line_set_Base::IsSet(), CBlast_def_line_Base::IsSetTaxid(), CBlast_def_line_Base::IsSetTitle(), ITERATE, kEmptyStr, CRef< C, Locker >::NotEmpty(), CDisplaySeqalign::PrepareBlastUngappedSeqalign(), s_SeqAlignSetToXMLHsps(), SSeqDBTaxInfo::scientific_name, CSeq_id::Score(), and ZERO_TAX_ID.

Referenced by s_SetBlastXMlHitList().

◆ s_SetBlastXMlHitList()

static void s_SetBlastXMlHitList	(	list< CRef< blastxml2::CHit > > &	hits,
		const IBlastXML2ReportData *	data,
		int	num
	)

static

Fills the list of blastxml2::CHit objects, given a list of Seq-aligns.

Parameters

hits	List of blastxml2::CHit objects to fill [in] [out]

Definition at line 460 of file blastxml2_format.cpp.

References data, CSeq_id::Match(), s_GetSubjectId(), s_SeqAlignToXMLHit(), CSeq_align_set_Base::Set(), and CSeq_align_Base::SetSegs().

Referenced by s_SetBlastXMLSearch().

◆ s_SetBlastXMLParameters()

static void s_SetBlastXMLParameters	(	blastxml2::CParameters &	params,
		const IBlastXML2ReportData *	data
	)

static

Fills the parameters part of the BLAST XML output.

Parameters

bxmlout	BLAST XML output object [in] [out]
data	Data structure, from which all necessary information can be retrieved [in]

Definition at line 515 of file blastxml2_format.cpp.

References data, NcbiEmptyString, str(), and val.

Referenced by s_FillBlastOutput().

◆ s_SetBlastXMLSearch()

static void s_SetBlastXMLSearch	(	blastxml2::CSearch &	search,
		const IBlastXML2ReportData *	data,
		int	num
	)

static

Definition at line 579 of file blastxml2_format.cpp.

References CSeq_id::AsFastaString(), data, eNa_strand_minus, CBioseq_Handle::GetBioseqCore(), CScope::GetBioseqHandle(), CSeqLocInfo::GetInterval(), GetLength(), GetSeqIdString(), CSeqLocInfo::GetStrand(), ITERATE, kEmptyStr, s_SetBlastXMlHitList(), s_SetBlastXMLStatistics(), and stats.

Referenced by s_FillBlastOutput().

◆ s_SetBlastXMLStatistics()

static void s_SetBlastXMLStatistics	(	blastxml2::CStatistics &	stats,
		const IBlastXML2ReportData *	data,
		int	num
	)

static

Fills the search statistics part of the BLAST XML output for all queries.

Parameters

stat_vec	Vector of the blastxml2::CStatics objects, to be filled. [in] [out]
data	Data structure, from which all necessary information can be retrieved [in]

Definition at line 562 of file blastxml2_format.cpp.

References data, and stats.

Referenced by s_SetBlastXMLSearch().