Implementation for the CSeqDBVol class, which provides an interface for all functionality of one database volume. More...

#include <ncbi_pch.hpp>
#include <objtools/blast/seqdb_reader/impl/seqdbvol.hpp>
#include "seqdboidlist.hpp"
#include <objects/general/general__.hpp>
#include <objects/seqfeat/seqfeat__.hpp>
#include <serial/objistr.hpp>
#include <serial/objostr.hpp>
#include <serial/objistrasnb.hpp>
#include <serial/objostrasnb.hpp>
#include <serial/serial.hpp>
#include <corelib/ncbimtx.hpp>
#include <sstream>

Include dependency graph for seqdbvol.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Typedefs
typedef set< pair< int, int > >	TRangeVector
	List of offset ranges as begin/end pairs. More...

Functions
static vector< Uint1 >	s_SeqDBMapNA2ToNA4Setup ()
	Build NA2 to NcbiNA4 translation table. More...

static void	s_SeqDBMapNA2ToNA4 (const char *buf2bit, vector< char > &buf4bit, int base_length)
	Convert sequence data from NA2 to NA4 format. More...

static vector< Uint1 >	s_SeqDBMapNA2ToNA8Setup ()
	Build NA2 to Ncbi-NA8 translation table. More...

static void	s_SeqDBMapNA2ToNA8 (const char buf2bit, char buf8bit, const SSeqDBSlice &range)
	Convert sequence data from NA2 to NA8 format. More...

static void	s_SeqDBMapNcbiNA8ToBlastNA8 (char *buf, const SSeqDBSlice &range)
	Convert sequence data from Ncbi-NA8 to Blast-NA8 format. More...

Uint4	s_ResLenNew (const vector< Int4 > &ambchars, Uint4 i)
	Get length of ambiguous region (new version) More...

Uint4	s_ResPosNew (const vector< Int4 > &ambchars, Uint4 i)
	Get position of ambiguous region (new version) More...

Uint4	s_ResVal (const vector< Int4 > &ambchars, Uint4 i)
	Get ambiguous residue value (old version) More...

Uint4	s_ResLenOld (const vector< Int4 > &ambchars, Uint4 i)
	Get ambiguous region length (old version) More...

Uint4	s_ResPosOld (const vector< Int4 > &ambchars, Uint4 i)
	Get ambiguous residue value (old version) More...

static void	s_SeqDBRebuildDNA_NA4 (vector< char > &buf4bit, const vector< Int4 > &amb_chars)
	Rebuild an ambiguous region from sequence and ambiguity data. More...

static void	s_SeqDBRebuildDNA_NA8 (char *seq, const vector< Int4 > &amb_chars, const SSeqDBSlice &region)
	Rebuild an ambiguous region from sequence and ambiguity data. More...

static void	s_SeqDBWriteSeqDataProt (CSeq_inst &seqinst, const char *seq_buffer, int length)
	Store protein sequence data in a Seq-inst. More...

static void	s_SeqDBWriteSeqDataNucl (CSeq_inst &seqinst, const char *seq_buffer, int length)
	Store non-ambiguous nucleotide sequence data in a Seq-inst. More...

static void	s_SeqDBWriteSeqDataNucl (CSeq_inst &seqinst, const char *seq_buffer, int length, vector< Int4 > &amb_chars)
	Store non-ambiguous nucleotide sequence data in a Seq-inst. More...

static void	s_GetBioseqTitle (CRef< CBlast_def_line_set > deflines, string &title)
	Get the title string for a CBioseq. More...

static bool	s_SeqDB_SeqIdIn (const list< CRef< CSeq_id > > &seqids, const CSeq_id &target)
	Search for a Seq-id in a list of Seq-ids. More...

static CRef< CBlast_def_line_set >	s_OssToDefline (const CUser_field::TData::TOss &oss)
	Efficiently decode a Blast-def-line-set from binary ASN.1. More...

template<class T >
CRef< CBlast_def_line_set >	s_ExtractBlastDefline (const T &bioseq)

static void	s_SeqDBMaskSequence (char seq, CSeqDB::TSequenceRanges masks, char mask_letter, const SSeqDBSlice &range)

void	SeqDB_UnpackAmbiguities (const CTempString &sequence, const CTempString &ambiguities, string &result)
	Unpack an ambiguous nucleotide sequence. More...

bool	s_IncludeDefline_Taxid (const CBlast_def_line &def, const set< TTaxId > &user_tax_ids)

bool	s_IncludeDefline_MaskFilter (const CBlast_def_line &def, Uint8 mask)

bool	s_IncludeDefline_NegativeTaxid (const CBlast_def_line &def, const set< TTaxId > &user_tax_ids)

template<class T >
static void	s_SeqDBFitsInFour (T id)

Variables
unsigned	SeqDB_ncbina8_to_blastna8 []

Detailed Description

Implementation for the CSeqDBVol class, which provides an interface for all functionality of one database volume.

Definition in file seqdbvol.cpp.

Typedef Documentation

◆ TRangeVector

typedef set< pair<int, int> > TRangeVector

List of offset ranges as begin/end pairs.

Definition at line 1587 of file seqdbvol.cpp.

Function Documentation

◆ s_ExtractBlastDefline()

template<class T >

CRef<CBlast_def_line_set> s_ExtractBlastDefline ( const T & bioseq )

Definition at line 1215 of file seqdbvol.cpp.

References _ASSERT, failure, CUser_object_Base::GetData(), CObject_id_Base::GetStr(), CUser_object_Base::GetType(), CObject_id_Base::IsStr(), ITERATE, kAsnDeflineObjLabel, and s_OssToDefline().

Referenced by CSeqDB::ExtractBlastDefline().

◆ s_GetBioseqTitle()

static void s_GetBioseqTitle	(	CRef< CBlast_def_line_set >	deflines,
		string &	title
	)

static

Get the title string for a CBioseq.

GetBioseq will use this function to get a title field when constructing the CBioseq object.

Parameters

deflines	The set of deflines for this sequence. [in]
title	The returned title string. [out]

Definition at line 938 of file seqdbvol.cpp.

References CBlast_def_line_Base::CanGetSeqid(), CBlast_def_line_Base::CanGetTitle(), CBlast_def_line_set_Base::Get(), CBlast_def_line_Base::GetSeqid(), and CBlast_def_line_Base::GetTitle().

Referenced by CSeqDBVol::GetBioseq().

◆ s_IncludeDefline_MaskFilter()

bool s_IncludeDefline_MaskFilter	(	const CBlast_def_line &	def,
		Uint8	mask
	)

Definition at line 1925 of file seqdbvol.cpp.

References CSeq_id::fAcc_predicted, fExcludeModel, CBlast_def_line_Base::GetSeqid(), ITERATE, and mask.

Referenced by CSeqDBVol::x_GetFilteredHeader().

◆ s_IncludeDefline_NegativeTaxid()

bool s_IncludeDefline_NegativeTaxid	(	const CBlast_def_line &	def,
		const set< TTaxId > &	user_tax_ids
	)

Definition at line 1938 of file seqdbvol.cpp.

References set< Key, Compare >::end(), set< Key, Compare >::find(), CBlast_def_line::GetTaxIds(), ITERATE, and set< Key, Compare >::size().

Referenced by CSeqDBVol::x_GetFilteredHeader().

◆ s_IncludeDefline_Taxid()

bool s_IncludeDefline_Taxid	(	const CBlast_def_line &	def,
		const set< TTaxId > &	user_tax_ids
	)

Definition at line 1892 of file seqdbvol.cpp.

References set< Key, Compare >::end(), set< Key, Compare >::find(), CBlast_def_line_Base::GetLinks(), CBlast_def_line_Base::GetTaxid(), set< Key, Compare >::insert(), CBlast_def_line_Base::IsSetLinks(), CBlast_def_line_Base::IsSetTaxid(), ITERATE, set< Key, Compare >::size(), and TAX_ID_FROM.

Referenced by CSeqDBVol::x_GetFilteredHeader().

◆ s_OssToDefline()

static CRef<CBlast_def_line_set> s_OssToDefline ( const CUser_field::TData::TOss & oss )

static

Efficiently decode a Blast-def-line-set from binary ASN.1.

Parameters

oss	Octet string sequence of binary ASN.1 data.
bdls	Blast def line set decoded from oss.

Definition at line 1176 of file seqdbvol.cpp.

References data, ITERATE, NULL, and ncbi::grid::netcache::search::fields::size.

Referenced by s_ExtractBlastDefline().

◆ s_ResLenNew()

Uint4 s_ResLenNew	(	const vector< Int4 > &	ambchars,
		Uint4	i
	)

inline

Get length of ambiguous region (new version)

Given an ambiguity element in the new format, this returns the length of the ambiguous region.

Parameters

ambchars	The packed ambiguity data. [in]
i	The index into the ambiguity data. [in]

Returns: The region length.

Definition at line 623 of file seqdbvol.cpp.

References i.

Referenced by s_SeqDBRebuildDNA_NA4(), and s_SeqDBRebuildDNA_NA8().

◆ s_ResLenOld()

Uint4 s_ResLenOld	(	const vector< Int4 > &	ambchars,
		Uint4	i
	)

inline

Get ambiguous region length (old version)

Given an ambiguity element in the old format, this returns the length of the ambiguous region.

Parameters

ambchars	The packed ambiguity data. [in]
i	The index into the ambiguity data. [in]

Returns: The residue value.

Definition at line 675 of file seqdbvol.cpp.

References i.

Referenced by s_SeqDBRebuildDNA_NA4(), and s_SeqDBRebuildDNA_NA8().

◆ s_ResPosNew()

Uint4 s_ResPosNew	(	const vector< Int4 > &	ambchars,
		Uint4	i
	)

inline

Get position of ambiguous region (new version)

Given an ambiguity element in the new format, this returns the position of the ambiguous region.

Parameters

ambchars	The packed ambiguity data. [in]
i	The index into the ambiguity data. [in]

Returns: The region length.

Definition at line 639 of file seqdbvol.cpp.

References i.

Referenced by s_SeqDBRebuildDNA_NA4(), and s_SeqDBRebuildDNA_NA8().

◆ s_ResPosOld()

Uint4 s_ResPosOld	(	const vector< Int4 > &	ambchars,
		Uint4	i
	)

inline

Get ambiguous residue value (old version)

Given an ambiguity element in the old format, this returns the position of the ambiguous region.

Parameters

ambchars	The packed ambiguity data. [in]
i	The index into the ambiguity data. [in]

Returns: The residue value.

Definition at line 691 of file seqdbvol.cpp.

References i.

Referenced by s_SeqDBRebuildDNA_NA4(), and s_SeqDBRebuildDNA_NA8().

◆ s_ResVal()

Uint4 s_ResVal	(	const vector< Int4 > &	ambchars,
		Uint4	i
	)

inline

Get ambiguous residue value (old version)

Given an ambiguity element in the old format, this returns the residue value to use for all bases in the ambiguous region.

Parameters

ambchars	The packed ambiguity data. [in]
i	The index into the ambiguity data. [in]

Returns: The residue value.

Definition at line 659 of file seqdbvol.cpp.

References i.

Referenced by s_SeqDBRebuildDNA_NA4(), and s_SeqDBRebuildDNA_NA8().

◆ s_SeqDB_SeqIdIn()

static bool s_SeqDB_SeqIdIn	(	const list< CRef< CSeq_id > > &	seqids,
		const CSeq_id &	target
	)

static

Search for a Seq-id in a list of Seq-ids.

This iterates over a list of Seq-ids, and returns true if a specific Seq-id is equivalent to one found in the list.

Parameters

seqids	A list of Seq-ids to search. [in]
target	The Seq-id to search for. [in]

Returns: True if the Seq-id was found.

Definition at line 1013 of file seqdbvol.cpp.

References CSeq_id::e_NO, CSeq_id::e_YES, and ITERATE.

Referenced by CSeqDBVol::GetBioseq(), and CSeqDBVol::x_GetTaxDefline().

◆ s_SeqDBFitsInFour()

template<class T >

static void s_SeqDBFitsInFour ( T id )

static

Definition at line 3059 of file seqdbvol.cpp.

References NCBI_THROW, and T.

Referenced by CSeqDBVol::GetGiBounds(), and CSeqDBVol::GetPigBounds().

◆ s_SeqDBMapNA2ToNA4()

static void s_SeqDBMapNA2ToNA4	(	const char *	buf2bit,
		vector< char > &	buf4bit,
		int	base_length
	)

static

Convert sequence data from NA2 to NA4 format.

This uses a translation table to convert nucleotide data. The input data is in NA2 format, the output data will be in NcbiNA4 format.

Parameters

buf2bit	The NA2 input data. [in]
buf4bit	The NcbiNA4 output data. [out]
base_length	The length (in bases) of the input data. [in]

Definition at line 389 of file seqdbvol.cpp.

References _ASSERT, base_length, i, and s_SeqDBMapNA2ToNA4Setup().

Referenced by s_SeqDBWriteSeqDataNucl().

◆ s_SeqDBMapNA2ToNA4Setup()

static vector<Uint1> s_SeqDBMapNA2ToNA4Setup ( )

static

Build NA2 to NcbiNA4 translation table.

This builds a translation table for nucleotide data. The table will be used by s_SeqDBMapNA2ToNA4(). The table is indexed by the packed nucleotide representation, or "NA2" format, which encodes four bases per byte. The elements of the table are the unpacked "Ncbi-NA4" representation, which encodes two bases per byte.

Returns: The NA2 to NA4 translation table

Definition at line 351 of file seqdbvol.cpp.

References convert().

Referenced by s_SeqDBMapNA2ToNA4().

◆ s_SeqDBMapNA2ToNA8()

static void s_SeqDBMapNA2ToNA8	(	const char *	buf2bit,
		char *	buf8bit,
		const SSeqDBSlice &	range
	)

static

Convert sequence data from NA2 to NA8 format.

This uses a translation table to convert nucleotide data. The input data is in NA2 format, the output data will be in Ncbi-NA8 format. This function also optionally adds sentinel bytes to the start and end of the data (needed by some applications).

Parameters

buf2bit	The NA2 input data. [in]
buf8bit	The start of the Ncbi-NA8 output data. [out]
buf8bit_end	The end of the Ncbi-NA8 output data. [out]
sentinel_bytes	Specify true if sentinel bytes should be included. [in]
range	The subregion of the sequence to work on. [in]

Definition at line 481 of file seqdbvol.cpp.

References _ASSERT, compile_time_bits::range(), and s_SeqDBMapNA2ToNA8Setup().

Referenced by CSeqDBVol::GetAmbigPartialSeq(), SeqDB_UnpackAmbiguities(), and CSeqDBVol::x_GetAmbigSeq().

◆ s_SeqDBMapNA2ToNA8Setup()

static vector<Uint1> s_SeqDBMapNA2ToNA8Setup ( )

static

Build NA2 to Ncbi-NA8 translation table.

This builds a translation table for nucleotide data. The table will be used by s_SeqDBMapNA2ToNA8(). The table is indexed by the packed nucleotide representation, or "NA2" format, which encodes four bases per byte. The elements of the table are the unpacked "Ncbi-NA8" representation, which encodes one base per byte.

Returns: The NA2 to NA8 translation table

Definition at line 440 of file seqdbvol.cpp.

References i.

Referenced by s_SeqDBMapNA2ToNA8().

◆ s_SeqDBMapNcbiNA8ToBlastNA8()

static void s_SeqDBMapNcbiNA8ToBlastNA8	(	char *	buf,
		const SSeqDBSlice &	range
	)

static

Convert sequence data from Ncbi-NA8 to Blast-NA8 format.

This uses a translation table to convert nucleotide data. The input data is in Ncbi-NA8 format, the output data will be in Blast-NA8 format. The data is converted in-place.

Parameters

buf	The array of nucleotides to convert. [in\|out]
range	The range of opearation. [in]

Definition at line 601 of file seqdbvol.cpp.

References buf, i, compile_time_bits::range(), and SeqDB_ncbina8_to_blastna8.

Referenced by CSeqDBVol::GetAmbigPartialSeq(), and CSeqDBVol::x_GetAmbigSeq().

◆ s_SeqDBMaskSequence()

static void s_SeqDBMaskSequence	(	char *	seq,
		CSeqDB::TSequenceRanges *	masks,
		char	mask_letter,
		const SSeqDBSlice &	range
	)

static

Definition at line 1499 of file seqdbvol.cpp.

References CSeqDB::TSequenceRanges::empty(), first(), i, max(), min(), compile_time_bits::range(), and ncbi::grid::netcache::search::fields::size.

Referenced by CSeqDBVol::GetAmbigPartialSeq(), and CSeqDBVol::x_GetAmbigSeq().

◆ s_SeqDBRebuildDNA_NA4()

static void s_SeqDBRebuildDNA_NA4	(	vector< char > &	buf4bit,
		const vector< Int4 > &	amb_chars
	)

static

Rebuild an ambiguous region from sequence and ambiguity data.

When sequence data for a blast database is built, ambiguous regions are replaced with random strings of the four standard nucleotides. The ambiguity data is seperately encoded as a sequence of integer values. This function unpacks the ambiguity data and replaces the randomized bases with correct (ambiguous) encodings. This version works with 4 bit representations.

Parameters

buf4bit	Sequence data for a sequence. [in\|out]
amb_chars	Corresponding ambiguous data. [in]

Definition at line 710 of file seqdbvol.cpp.

References i, s_ResLenNew(), s_ResLenOld(), s_ResPosNew(), s_ResPosOld(), and s_ResVal().

Referenced by s_SeqDBWriteSeqDataNucl().

◆ s_SeqDBRebuildDNA_NA8()

static void s_SeqDBRebuildDNA_NA8	(	char *	seq,
		const vector< Int4 > &	amb_chars,
		const SSeqDBSlice &	region
	)

static

Rebuild an ambiguous region from sequence and ambiguity data.

When sequence data for a blast database is built, ambiguous regions are replaced with random strings of the four standard nucleotides. The ambiguity data is seperately encoded as a sequence of integer values. This function unpacks the ambiguity data and replaces the randomized bases with correct (ambiguous) encodings. This version works with 8 bit representations.

Parameters

seq	Sequence data for a sequence. [in\|out]
amb_chars	Corresponding ambiguous data. [in]
region	If non-null, the part of the sequence to get. [in]

Definition at line 785 of file seqdbvol.cpp.

References SSeqDBSlice::begin, SSeqDBSlice::end, i, s_ResLenNew(), s_ResLenOld(), s_ResPosNew(), s_ResPosOld(), and s_ResVal().

Referenced by CSeqDBVol::GetAmbigPartialSeq(), SeqDB_UnpackAmbiguities(), and CSeqDBVol::x_GetAmbigSeq().

◆ s_SeqDBWriteSeqDataNucl() [1/2]

static void s_SeqDBWriteSeqDataNucl	(	CSeq_inst &	seqinst,
		const char *	seq_buffer,
		int	length
	)

static

Store non-ambiguous nucleotide sequence data in a Seq-inst.

This function reads length elements from seq_buffer and stores them in a Seq-inst object. It also sets appropriate encoding information in that object. No ambiguity information is used. The input array is assumed to be in 2 bit representation.

Parameters

seqinst	The Seq-inst to return the data in. [out]
seq_buffer	The input sequence data. [in]
length	The length (in bases) of the input data. [in]

Definition at line 877 of file seqdbvol.cpp.

References CSeq_inst_Base::eMol_na, i, CSeq_inst_Base::SetMol(), and CSeq_inst_Base::SetSeq_data().

Referenced by CSeqDBVol::GetBioseq().

◆ s_SeqDBWriteSeqDataNucl() [2/2]

static void s_SeqDBWriteSeqDataNucl	(	CSeq_inst &	seqinst,
		const char *	seq_buffer,
		int	length,
		vector< Int4 > &	amb_chars
	)

static

Store non-ambiguous nucleotide sequence data in a Seq-inst.

This function reads length elements from seq_buffer and stores them in a Seq-inst object. It also sets appropriate encoding information in that object. No ambiguity information is used. The input array is assumed to be in Ncbi-NA4 representation.

Parameters

seqinst	The Seq-inst to return the data in. [out]
seq_buffer	The input sequence data in Ncbi-NA4 format. [in]
length	The length (in bases) of the input data. [in]
amb_chars	The ambiguity data for this sequence. [in]

Definition at line 915 of file seqdbvol.cpp.

References CSeq_inst_Base::eMol_na, s_SeqDBMapNA2ToNA4(), s_SeqDBRebuildDNA_NA4(), CSeq_inst_Base::SetMol(), and CSeq_inst_Base::SetSeq_data().

◆ s_SeqDBWriteSeqDataProt()

static void s_SeqDBWriteSeqDataProt	(	CSeq_inst &	seqinst,
		const char *	seq_buffer,
		int	length
	)

static

Store protein sequence data in a Seq-inst.

This function reads length elements from seq_buffer and stores them in a Seq-inst object. It also sets appropriate encoding information in that object.

Parameters

seqinst	The Seq-inst to return the data in. [out]
seq_buffer	The input sequence data. [in]
length	The length (in bases) of the input data. [in]

Definition at line 840 of file seqdbvol.cpp.

References CSeq_inst_Base::eMol_aa, i, CSeq_inst_Base::SetMol(), and CSeq_inst_Base::SetSeq_data().

Referenced by CSeqDBVol::GetBioseq().

◆ SeqDB_UnpackAmbiguities()

void SeqDB_UnpackAmbiguities	(	const CTempString &	sequence,
		const CTempString &	ambiguities,
		string &	result
	)

Unpack an ambiguous nucleotide sequence.

This method provides a way to unpack nucleotide sequence data that has been packed in blast database format. One source of such data is the GetRawSeqAndAmbig() method in the CSeqDBExpert class. The output format is ncbi8na.

Parameters

sequence	Sequence data in NA2 format with encoded length. [in]
ambiguities	Sequence ambiguities packed in blastdb format. [in]
result	Unpacked sequence in Ncbi NA8 format. [out]

Definition at line 1698 of file seqdbvol.cpp.

References A, base_length, CTempString::data(), free(), i, CTempString::length(), malloc(), NCBI_THROW, compile_time_bits::range(), result, s_SeqDBMapNA2ToNA8(), s_SeqDBRebuildDNA_NA8(), and SeqDB_GetStdOrd().

Referenced by CWriteDB_Impl::x_ComputeHash().

Variable Documentation

◆ SeqDB_ncbina8_to_blastna8

unsigned SeqDB_ncbina8_to_blastna8[]

Initial value:

Definition at line 571 of file seqdbvol.cpp.

Referenced by s_SeqDBMapNcbiNA8ToBlastNA8().

Typedefs

Functions

Variables

Detailed Description

Typedef Documentation

◆ TRangeVector

Function Documentation

◆ s_ExtractBlastDefline()

◆ s_GetBioseqTitle()

◆ s_IncludeDefline_MaskFilter()

◆ s_IncludeDefline_NegativeTaxid()

◆ s_IncludeDefline_Taxid()

◆ s_OssToDefline()

◆ s_ResLenNew()

◆ s_ResLenOld()

◆ s_ResPosNew()

◆ s_ResPosOld()

◆ s_ResVal()

◆ s_SeqDB_SeqIdIn()

◆ s_SeqDBFitsInFour()

◆ s_SeqDBMapNA2ToNA4()

◆ s_SeqDBMapNA2ToNA4Setup()

◆ s_SeqDBMapNA2ToNA8()

◆ s_SeqDBMapNA2ToNA8Setup()

◆ s_SeqDBMapNcbiNA8ToBlastNA8()

◆ s_SeqDBMaskSequence()

◆ s_SeqDBRebuildDNA_NA4()

◆ s_SeqDBRebuildDNA_NA8()

◆ s_SeqDBWriteSeqDataNucl() [1/2]

◆ s_SeqDBWriteSeqDataNucl() [2/2]

◆ s_SeqDBWriteSeqDataProt()

◆ SeqDB_UnpackAmbiguities()

Variable Documentation

◆ SeqDB_ncbina8_to_blastna8