blast_app_util.cpp File Reference

Utility functions for BLAST command line applications. More...

#include <ncbi_pch.hpp>
#include "blast_app_util.hpp"
#include <serial/serial.hpp>
#include <serial/objostr.hpp>
#include <objtools/data_loaders/blastdb/bdbloader.hpp>
#include <algo/blast/api/remote_blast.hpp>
#include <algo/blast/api/objmgr_query_data.hpp>
#include <algo/blast/api/blast_options_builder.hpp>
#include <algo/blast/api/search_strategy.hpp>
#include <algo/blast/blastinput/blast_input.hpp>
#include <algo/blast/blastinput/psiblast_args.hpp>
#include <algo/blast/blastinput/tblastn_args.hpp>
#include <algo/blast/blastinput/blast_scope_src.hpp>
#include <objmgr/util/sequence.hpp>
#include <objects/scoremat/Pssm.hpp>
#include <serial/typeinfo.hpp>
#include <objtools/data_loaders/blastdb/bdbloader_rmt.hpp>
#include <algo/blast/format/blast_format.hpp>
#include <objtools/align_format/format_flags.hpp>

Include dependency graph for blast_app_util.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Functions
	USING_SCOPE (objects)
	USING_SCOPE (blast)
CRef< blast::CRemoteBlast >	InitializeRemoteBlast (CRef< blast::IQueryFactory > queries, CRef< blast::CBlastDatabaseArgs > db_args, CRef< blast::CBlastOptionsHandle > opts_hndl, bool verbose_output, const string &client_id, CRef< objects::CPssmWithParameters > pssm)
	Initializes a CRemoteBlast instance for usage by command line BLAST binaries. More...
blast::SDataLoaderConfig	InitializeQueryDataLoaderConfiguration (bool query_is_protein, CRef< blast::CLocalDbAdapter > db_adapter)
	Initialize the data loader configuration for the query. More...
void	InitializeSubject (CRef< blast::CBlastDatabaseArgs > db_args, CRef< blast::CBlastOptionsHandle > opts_hndl, bool is_remote_search, CRef< blast::CLocalDbAdapter > &db_adapter, CRef< objects::CScope > &scope)
	Initializes the subject/database as well as its scope. More...
string	RegisterOMDataLoader (CRef< CSeqDB > db_handle)
	Register the BLAST database data loader using the already initialized CSeqDB object. More...
static CRef< blast::CExportStrategy >	s_InitializeExportStrategy (CRef< blast::IQueryFactory > queries, CRef< blast::CBlastDatabaseArgs > db_args, CRef< blast::CBlastOptionsHandle > opts_hndl, const string &client_id, CRef< objects::CPssmWithParameters > pssm, unsigned int num_iters)
static void	s_ExportSearchStrategy (CNcbiOstream *out, CRef< blast::IQueryFactory > queries, CRef< blast::CBlastOptionsHandle > options_handle, CRef< blast::CBlastDatabaseArgs > db_args, CRef< objects::CPssmWithParameters > pssm, unsigned int num_iters)
	Real implementation of search strategy extraction. More...
static TSeqLocVector	s_ConvertBioseqs2TSeqLocVector (const CBlast4_subject::TSequences &subjects)
	Converts a list of Bioseqs into a TSeqLocVector. More...
static void	s_ImportPssm (const CBlast4_queries &queries, CRef< blast::CBlastOptionsHandle > opts_hndl, blast::CBlastAppArgs *cmdline_args)
	Import PSSM into the command line arguments object. More...
static void	s_ImportQueries (const CBlast4_queries &queries, CRef< blast::CBlastOptionsHandle > opts_hndl, blast::CBlastAppArgs *cmdline_args)
	Import queries into command line arguments object. More...
static CRef< blast::CBlastDatabaseArgs >	s_ImportDatabase (const CBlast4_subject &subj, CBlastOptionsBuilder &opts_builder, bool subject_is_protein, bool is_remote_search)
	Import the database and return it in a CBlastDatabaseArgs object. More...
static CRef< blast::CBlastDatabaseArgs >	s_ImportSubjects (const CBlast4_subject &subj, bool subject_is_protein)
	Import the subject sequences into a CBlastDatabaseArgs object. More...
static void	s_ImportSearchStrategy (CNcbiIstream *in, blast::CBlastAppArgs *cmdline_args, bool is_remote_search, bool override_query, bool override_subject)
	Imports search strategy, using CImportStrategy. More...
bool	RecoverSearchStrategy (const CArgs &args, blast::CBlastAppArgs *cmdline_args)
	Recover search strategy from input file. More...
void	SaveSearchStrategy (const CArgs &args, blast::CBlastAppArgs *cmdline_args, CRef< blast::IQueryFactory > queries, CRef< blast::CBlastOptionsHandle > opts_hndl, CRef< objects::CPssmWithParameters > pssm, unsigned int num_iters)
	Save the search strategy corresponding to the current command line search. More...
static void	s_ExtractSeqidsAndRanges (const blast::CSearchResultSet &results, CScope::TIds &ids, vector< TSeqRange > &ranges)
	Extracts the subject sequence IDs and ranges from the BLAST results. More...
static bool	s_IsUsingRemoteBlastDbDataLoader ()
	Returns true if the remote BLAST DB data loader is being used. More...
static bool	s_IsPrefetchFormat (blast::CFormattingArgs::EOutputFormat format_type)
static bool	s_PreFetchSeqs (const blast::CSearchResultSet &results, blast::CFormattingArgs::EOutputFormat format_type)
void	BlastFormatter_PreFetchSequenceData (const blast::CSearchResultSet &results, CRef< CScope > scope, blast::CFormattingArgs::EOutputFormat format_type)
	This method optimize the retrieval of sequence data to scope. More...
CRef< CBlastAncillaryData >	ExtractPssmAncillaryData (const CPssmWithParameters &pssm)
	Auxiliary function to extract the ancillary data from the PSSM. More...
void	CheckForFreqRatioFile (const string &rps_dbname, CRef< CBlastOptionsHandle > &opt_handle, bool isRpsblast)
bool	IsIStreamEmpty (CNcbiIstream &in)
string	GetCmdlineArgs (const CNcbiArguments &a)
bool	UseXInclude (const CFormattingArgs &f, const string &s)
string	GetSubjectFile (const CArgs &args)
	Get name of subject file @parameter args arguments class [in]. More...
void	PrintErrorArchive (const CArgs &a, const list< CRef< CBlast4_error > > &msg)
	Function to print blast archive with only error messages (search failed) to output stream. More...
void	QueryBatchCleanup ()
	Clean up formatter scope and release. More...
void	LogQueryInfo (CBlastUsageReport &report, const CBlastInput &q_info)
void	LogBlastOptions (blast::CBlastUsageReport &report, const CBlastOptions &opt)
void	LogCmdOptions (blast::CBlastUsageReport &report, const CBlastAppArgs &args)
int	GetMTByQueriesBatchSize (EProgram p, int num_threads, const string &task)
void	MTByQueries_DBSize_Warning (const Int8 length_limit, bool is_db_protein)
void	CheckMTByQueries_QuerySize (EProgram prog, int batch_size)

Detailed Description

Utility functions for BLAST command line applications.

Definition in file blast_app_util.cpp.

Function Documentation

BlastFormatter_PreFetchSequenceData()

void BlastFormatter_PreFetchSequenceData	(	const blast::CSearchResultSet &	results,
		CRef< CScope >	scope,
		blast::CFormattingArgs::EOutputFormat	format_type
	)

This method optimize the retrieval of sequence data to scope.

Parameters

results	BLAST results [in]
scope	CScope object from which the sequence data will be fetched [in]
format_type	Blast output fomat type

Definition at line 756 of file blast_app_util.cpp.

References _ASSERT, _TRACE, ERR_POST, LoadSequencesToScope(), CRef< C, Locker >::NotEmpty(), results, s_ExtractSeqidsAndRanges(), s_IsUsingRemoteBlastDbDataLoader(), s_PreFetchSeqs(), and Warning().

Referenced by CPsiBlastApp::DoIterations(), CDeltaBlastApp::DoPsiBlastIterations(), CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CIgBlastnApp::CIgFormatter::Main(), CBlastFormatterApp::PrintFormattedOutput(), CDeltaBlastApp::Run(), CTblastxApp::Run(), CIgBlastpApp::Run(), CPsiBlastApp::x_RunLocalPsiBlastIterations(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

CheckForFreqRatioFile()

void CheckForFreqRatioFile	(	const string &	rps_dbname,
		CRef< CBlastOptionsHandle > &	opt_handle,
		bool	isRpsblast
	)

Definition at line 806 of file blast_app_util.cpp.

References eNoCompositionBasedStats, CSeqDB::eProtein, ERR_POST, f, CSeqDB::FindVolumePaths(), CBlastOptions::GetCompositionBasedStats(), CBlastOptionsHandle::GetOptions(), i, NStr::Join(), msg(), NCBI_RETHROW, CBlastOptions::SetCompositionBasedStats(), CBlastOptionsHandle::SetOptions(), NStr::Split(), and Warning().

Referenced by CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), and CRPSBlastApp::x_RunMTBySplitDB().

CheckMTByQueries_QuerySize()

void CheckMTByQueries_QuerySize	(	EProgram	prog,
		int	batch_size
	)

Definition at line 1013 of file blast_app_util.cpp.

References Blast_QueryIsProtein(), EProgramToEBlastProgramType(), ERR_POST, NStr::IntToString(), prog, and Warning().

Referenced by CRPSBlastApp::x_RunMTBySplitQuery(), and CRPSTBlastnApp::x_RunMTBySplitQuery().

ExtractPssmAncillaryData()

CRef<CBlastAncillaryData> ExtractPssmAncillaryData

(

const CPssmWithParameters &

pssm

)

Auxiliary function to extract the ancillary data from the PSSM.

Definition at line 791 of file blast_app_util.cpp.

References _ASSERT, CPssmWithParameters_Base::CanGetPssm(), CPssm::GetH(), CPssm::GetHUngapped(), CPssm::GetKappa(), CPssm::GetKappaUngapped(), CPssm::GetLambda(), CPssm::GetLambdaUngapped(), CPssmWithParameters_Base::GetPssm(), and lambda().

Referenced by CDeltaBlastApp::SavePssmToFile(), and CPsiBlastApp::SavePssmToFile().

GetCmdlineArgs()

string GetCmdlineArgs

(

const CNcbiArguments &

a

)

Definition at line 877 of file blast_app_util.cpp.

References a, cmd, i, and kEmptyStr.

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CBlastFormatterApp::PrintFormattedOutput(), CBlastFormatterVdbApp::PrintFormattedOutput(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CVDBBlastnApp::Run(), CVDBTblastnApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

GetMTByQueriesBatchSize()

int GetMTByQueriesBatchSize	(	EProgram	p,
		int	num_threads,
		const string &	task
	)

Definition at line 980 of file blast_app_util.cpp.

References GetQueryBatchSize(), and NStr::StringToInt().

Referenced by CBlastnApp::x_RunMTBySplitQuery(), CBlastpApp::x_RunMTBySplitQuery(), CBlastxApp::x_RunMTBySplitQuery(), and CTblastnApp::x_RunMTBySplitQuery().

GetSubjectFile()

string GetSubjectFile

(

const CArgs &

args

)

Get name of subject file @parameter args arguments class [in].

Returns

Name of subject file.

Definition at line 902 of file blast_app_util.cpp.

References CArgs::Exist(), and kArgSubject.

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CTblastnNode::Main(), CTblastxApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

InitializeQueryDataLoaderConfiguration()

blast::SDataLoaderConfig InitializeQueryDataLoaderConfiguration	(	bool	query_is_protein,
		CRef< blast::CLocalDbAdapter >	db_adapter
	)

Initialize the data loader configuration for the query.

Parameters

query_is_protein	is/are the query sequence(s) protein? [in]
db_adapter	the database/subject information [in]

Definition at line 128 of file blast_app_util.cpp.

References _TRACE, SDataLoaderConfig::m_BlastDbName, SDataLoaderConfig::m_UseBlastDbs, SDataLoaderConfig::m_UseGenbank, SDataLoaderConfig::OptimizeForWholeLargeSequenceRetrieval(), and NStr::Split().

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

InitializeRemoteBlast()

CRef<blast::CRemoteBlast> InitializeRemoteBlast	(	CRef< blast::IQueryFactory >	queries,
		CRef< blast::CBlastDatabaseArgs >	db_args,
		CRef< blast::CBlastOptionsHandle >	opts_hndl,
		bool	verbose_output,
		const string &	client_id = kEmptyStr,
		CRef< objects::CPssmWithParameters >	pssm = CRef< objects::CPssmWithParameters >()
	)

Initializes a CRemoteBlast instance for usage by command line BLAST binaries.

Parameters

queries	query sequence(s) or NULL in case of PSSM input [in]
db_args	database/subject arguments [in]
opts_hndl	BLAST options handle [in]
verbose_output	set to true if CRemoteBlast should produce verbose output [in]
pssm	PSSM to use for single iteration remote PSI-BLAST

Exceptions

CInputException

in case of remote PSI-BL2SEQ, as it's not supported

Definition at line 84 of file blast_app_util.cpp.

References _ASSERT, CRef< C, Locker >::Empty(), kEmptyStr, NCBI_THROW, CRef< C, Locker >::NotEmpty(), CRef< C, Locker >::Reset(), CRemoteBlast::SetClientId(), and CRemoteBlast::SetVerbose().

Referenced by CPsiBlastApp::DoIterations(), CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CDeltaBlastApp::Run(), CTblastxApp::Run(), CRMBlastnApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

InitializeSubject()

void InitializeSubject	(	CRef< blast::CBlastDatabaseArgs >	db_args,
		CRef< blast::CBlastOptionsHandle >	opts_hndl,
		bool	is_remote_search,
		CRef< blast::CLocalDbAdapter > &	db_adapter,
		CRef< objects::CScope > &	scope
	)

Initializes the subject/database as well as its scope.

Parameters

db_args	database/subject arguments [in]
opts_hndl	BLAST options handle [in]
is_remote_search	true if it's a remote search, otherwise false [in]
db_adapter	Database/subject adapter [out]
scope	subject scope [out]

Definition at line 163 of file blast_app_util.cpp.

References _ASSERT, _TRACE, CBlastScopeSource::AddDataLoaders(), Blast_SubjectIsProtein(), CRef< C, Locker >::Empty(), CSearchDatabase::GetDatabaseName(), CObjectManager::GetInstance(), CSearchDatabase::GetSeqDb(), kEmptyStr, CBlastDatabaseArgs::kSubjectsDataLoaderPriority, CBlastScopeSource::NewScope(), CRef< C, Locker >::NotEmpty(), RegisterOMDataLoader(), CRef< C, Locker >::Reset(), and true.

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CIgBlastnApp::CIgWorker::Main(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CIgBlastnApp::Run(), CIgBlastpApp::Run(), CRMBlastnApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

IsIStreamEmpty()

bool IsIStreamEmpty

(

CNcbiIstream &

in

)

Definition at line 845 of file blast_app_util.cpp.

References in().

Referenced by CBlastnNode::Main(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), CTblastnApp::x_RunMTBySplitDB(), CBlastnApp::x_RunMTBySplitQuery(), CBlastpApp::x_RunMTBySplitQuery(), CBlastxApp::x_RunMTBySplitQuery(), CRPSBlastApp::x_RunMTBySplitQuery(), CRPSTBlastnApp::x_RunMTBySplitQuery(), and CTblastnApp::x_RunMTBySplitQuery().

LogBlastOptions()

void LogBlastOptions	(	blast::CBlastUsageReport &	report,
		const CBlastOptions &	opt
	)

Definition at line 946 of file blast_app_util.cpp.

References Blast_ProgramIsNucleotide(), Blast_ProgramNameFromType(), CBlastUsageReport::eCompBasedStats, CBlastUsageReport::eEvalueThreshold, CBlastUsageReport::eHitListSize, CBlastUsageReport::eProgram, CBlastOptions::GetCompositionBasedStats(), CBlastOptions::GetEvalueThreshold(), CBlastOptions::GetHitlistSize(), and CBlastOptions::GetProgramType().

Referenced by CBlastnApp::x_RunMTBySplitQuery(), CBlastpApp::x_RunMTBySplitQuery(), CBlastxApp::x_RunMTBySplitQuery(), CRPSBlastApp::x_RunMTBySplitQuery(), CRPSTBlastnApp::x_RunMTBySplitQuery(), and CTblastnApp::x_RunMTBySplitQuery().

LogCmdOptions()

void LogCmdOptions	(	blast::CBlastUsageReport &	report,
		const CBlastAppArgs &	args
	)

Definition at line 957 of file blast_app_util.cpp.

References CBlastUsageReport::eDBDate, CBlastUsageReport::eDBLength, CBlastUsageReport::eDBName, CBlastUsageReport::eDBNumSeqs, CBlastUsageReport::eOutputFmt, CBlastAppArgs::GetBlastDatabaseArgs(), CSeqDB::GetDate(), CSeqDB::GetDBNameList(), CFormattingArgs::GetFormattedOutputChoice(), CBlastAppArgs::GetFormattingArgs(), CSeqDB::GetNumSeqs(), CDirEntry::GetPathSeparator(), CBlastDatabaseArgs::GetSearchDatabase(), CSearchDatabase::GetSeqDb(), CSeqDB::GetTotalLength(), and CRef< C, Locker >::NotEmpty().

Referenced by CBlastnApp::x_RunMTBySplitQuery(), CBlastpApp::x_RunMTBySplitQuery(), CBlastxApp::x_RunMTBySplitQuery(), CRPSBlastApp::x_RunMTBySplitQuery(), CRPSTBlastnApp::x_RunMTBySplitQuery(), and CTblastnApp::x_RunMTBySplitQuery().

LogQueryInfo()

void LogQueryInfo	(	CBlastUsageReport &	report,
		const CBlastInput &	q_info
	)

Definition at line 939 of file blast_app_util.cpp.

References CBlastUsageReport::AddParam(), CBlastUsageReport::eNumQueries, CBlastUsageReport::eTotalQueryLength, CBlastInput::GetNumSeqsProcessed(), and CBlastInput::GetTotalLengthProcessed().

Referenced by CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CVDBBlastnApp::Run(), CVDBTblastnApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

MTByQueries_DBSize_Warning()

void MTByQueries_DBSize_Warning	(	const Int8	length_limit,
		bool	is_db_protein
	)

Definition at line 998 of file blast_app_util.cpp.

References ERR_POST, NStr::fWithCommas, NStr::Int8ToString(), and Warning().

Referenced by CBlastnApp::Run(), CBlastpApp::Run(), CBlastxApp::Run(), and CTblastnApp::Run().

PrintErrorArchive()

void PrintErrorArchive	(	const CArgs &	a,
		const list< CRef< CBlast4_error > > &	msg
	)

Function to print blast archive with only error messages (search failed) to output stream.

Parameters

a	cmdline args [in]
msg	list of errors and warning to be added to blast4 archive for printing

Definition at line 912 of file blast_app_util.cpp.

References a, CFormattingArgs::eArchiveFormat, info, kArgOutput, kArgOutputFormat, msg(), CBlastFormat::PrintArchive(), results, CBlast4_request_Base::SetBody(), CBlast4_archive_Base::SetMessages(), CBlast4_archive_Base::SetRequest(), CBlast4_archive_Base::SetResults(), NStr::StringToInt(), and NStr::TruncateSpaces().

Referenced by CBlastnApp::Run(), CBlastpApp::Run(), CBlastxApp::Run(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastnApp::Run(), CTblastxApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), CTblastnApp::x_RunMTBySplitDB(), CBlastnApp::x_RunMTBySplitQuery(), CBlastpApp::x_RunMTBySplitQuery(), CBlastxApp::x_RunMTBySplitQuery(), CRPSBlastApp::x_RunMTBySplitQuery(), CRPSTBlastnApp::x_RunMTBySplitQuery(), and CTblastnApp::x_RunMTBySplitQuery().

QueryBatchCleanup()

void QueryBatchCleanup

(

)

Clean up formatter scope and release.

Definition at line 930 of file blast_app_util.cpp.

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

RecoverSearchStrategy()

bool RecoverSearchStrategy	(	const CArgs &	args,
		blast::CBlastAppArgs *	cmdline_args
	)

Recover search strategy from input file.

Parameters

args	the command line arguments provided by the application [in]
cmdline_args	output command line arguments. Will have the database arguments set, as well as options handle [in|out] @Return true if recovered from save search strategy

Definition at line 608 of file blast_app_util.cpp.

References ERR_POST, CArgs::Exist(), CBlastDatabaseArgs::HasBeenSet(), CMbIndexArgs::HasBeenSet(), in(), kArgQuery, kArgRemote, kArgUseIndex, kDfltArgQuery, kDfltArgUseIndex, s_ImportSearchStrategy(), and Warning().

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CBlastnApp::Run(), CBlastpApp::Run(), CBlastxApp::Run(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastnApp::Run(), CTblastxApp::Run(), CRMBlastnApp::Run(), CVDBBlastnApp::Run(), CVDBTblastnApp::Run(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitQuery(), and CRPSTBlastnApp::x_RunMTBySplitQuery().

RegisterOMDataLoader()

string RegisterOMDataLoader

(

CRef< CSeqDB >

db_handle

)

Register the BLAST database data loader using the already initialized CSeqDB object.

Parameters

db_handle

properly initialized CSeqDB instance [in]

Returns

name of the BLAST data data loader (to be added to the CScope object)

Definition at line 246 of file blast_app_util.cpp.

References _TRACE, CObjectManager::eDefault, CObjectManager::GetInstance(), CBlastDbDataLoader::GetLoaderNameFromArgs(), CBlastDatabaseArgs::kSubjectsDataLoaderPriority, om, and CBlastDbDataLoader::RegisterInObjectManager().

Referenced by InitializeSubject(), CDeltaBlastApp::Run(), CSeedTopApp::Run(), and s_InitializeSubject().

s_ConvertBioseqs2TSeqLocVector()

static TSeqLocVector s_ConvertBioseqs2TSeqLocVector ( const CBlast4_subject::TSequences &  subjects )

static

Converts a list of Bioseqs into a TSeqLocVector.

All Bioseqs are added to the same CScope object

Parameters

subjects

Bioseqs to convert

Definition at line 355 of file blast_app_util.cpp.

References CScope::AddBioseq(), CSeq_id::BestRank(), FindBestChoice(), CObjectManager::GetInstance(), and ITERATE.

Referenced by s_ImportSubjects().

s_ExportSearchStrategy()

static void s_ExportSearchStrategy ( CNcbiOstream *  out, CRef< blast::IQueryFactory >  queries, CRef< blast::CBlastOptionsHandle >  options_handle, CRef< blast::CBlastDatabaseArgs >  db_args, CRef< objects::CPssmWithParameters >  pssm, unsigned int  num_iters  )

static

Real implementation of search strategy extraction.

Definition at line 319 of file blast_app_util.cpp.

References _ASSERT, CBlastException::eNotSupported, CException::GetErrCode(), kEmptyStr, NCBI_THROW, out(), and s_InitializeExportStrategy().

Referenced by SaveSearchStrategy().

s_ExtractSeqidsAndRanges()

static void s_ExtractSeqidsAndRanges ( const blast::CSearchResultSet &  results, CScope::TIds &  ids, vector< TSeqRange > &  ranges  )

static

Extracts the subject sequence IDs and ranges from the BLAST results.

Note

if this ever needs to be refactored for popular developer consumption, this function should operate on CSeq_align_set as opposed to blast::CSearchResultSet

Definition at line 659 of file blast_app_util.cpp.

References _ASSERT, eNa_strand_minus, eNa_strand_plus, CSeq_id_Handle::GetHandle(), ITERATE, r(), compile_time_bits::range(), result, results, CRange_Base::SetFrom(), and COpenRange< Position >::SetToOpen().

Referenced by BlastFormatter_PreFetchSequenceData().

s_ImportDatabase()

static CRef<blast::CBlastDatabaseArgs> s_ImportDatabase ( const CBlast4_subject &  subj, CBlastOptionsBuilder &  opts_builder, bool  subject_is_protein, bool  is_remote_search  )

static

Import the database and return it in a CBlastDatabaseArgs object.

Definition at line 444 of file blast_app_util.cpp.

References _ASSERT, CSeqDBGiList::AddGi(), CSeqDBGiList::AddTaxIds(), dbname(), CSearchDatabase::eBlastDbIsNucleotide, CSearchDatabase::eBlastDbIsProtein, eSoftSubjMasking, CBlast4_subject_Base::GetDatabase(), CBlastOptionsBuilder::GetDbFilteringAlgorithmId(), CBlastOptionsBuilder::GetDbFilteringAlgorithmKey(), CBlastOptionsBuilder::GetEntrezQuery(), CBlastOptionsBuilder::GetGiList(), CBlastOptionsBuilder::GetNegativeGiList(), CBlastOptionsBuilder::GetNegativeTaxidList(), CBlastOptionsBuilder::GetSubjectMaskingType(), CBlastOptionsBuilder::GetTaxidList(), CBlastOptionsBuilder::HasDbFilteringAlgorithmId(), CBlastOptionsBuilder::HasDbFilteringAlgorithmKey(), CBlastOptionsBuilder::HasSubjectMaskingType(), CBlastOptionsBuilder::HaveEntrezQuery(), CBlastOptionsBuilder::HaveGiList(), CBlastOptionsBuilder::HaveNegativeGiList(), CBlastOptionsBuilder::HaveNegativeTaxidList(), CBlastOptionsBuilder::HaveTaxidList(), CBlast4_subject_Base::IsDatabase(), ITERATE, msg(), CSearchDatabase::SetEntrezQueryLimitation(), CSearchDatabase::SetFilteringAlgorithm(), CSearchDatabase::SetGiList(), CSearchDatabase::SetNegativeGiList(), CBlastDatabaseArgs::SetSearchDatabase(), and string.

Referenced by s_ImportSearchStrategy().

s_ImportPssm()

static void s_ImportPssm ( const CBlast4_queries &  queries, CRef< blast::CBlastOptionsHandle >  opts_hndl, blast::CBlastAppArgs *  cmdline_args  )

static

Import PSSM into the command line arguments object.

Definition at line 372 of file blast_app_util.cpp.

References Blast_ProgramNameFromType(), CBlast4_queries_Base::GetPssm(), msg(), NCBI_THROW, NULL, CPsiBlastAppArgs::SetInputPssm(), and CTblastnAppArgs::SetInputPssm().

Referenced by s_ImportSearchStrategy().

s_ImportQueries()

static void s_ImportQueries ( const CBlast4_queries &  queries, CRef< blast::CBlastOptionsHandle >  opts_hndl, blast::CBlastAppArgs *  cmdline_args  )

static

Import queries into command line arguments object.

Definition at line 396 of file blast_app_util.cpp.

References _ASSERT, CTmpFile::AsOutputFile(), Blast_QueryIsProtein(), CFastaOstream::eAssembleParts, CTmpFile::eIfExists_Throw, CTmpFile::eNoRemove, CBlast4_queries_Base::GetBioseq_set(), CScope::GetBioseqHandle(), CTmpFile::GetFileName(), CBlast4_queries_Base::GetSeq_loc_list(), CBioseq_set_Base::GetSeq_set(), CBlast4_queries_Base::IsBioseq_set(), CBlast4_queries_Base::IsSeq_loc_list(), ITERATE, CBlastScopeSource::NewScope(), SDataLoaderConfig::OptimizeForWholeLargeSequenceRetrieval(), out(), prog, CRef< C, Locker >::Reset(), and CBlastScopeSource::RevokeBlastDbDataLoader().

Referenced by s_ImportSearchStrategy().

s_ImportSearchStrategy()

static void s_ImportSearchStrategy ( CNcbiIstream *  in, blast::CBlastAppArgs *  cmdline_args, bool  is_remote_search, bool  override_query, bool  override_subject  )

static

Imports search strategy, using CImportStrategy.

Definition at line 533 of file blast_app_util.cpp.

References _ASSERT, _TRACE, Blast_SubjectIsProtein(), BlastNumber2Program(), ERR_POST, ExtractBlast4Request(), COpenRange< TSeqPos >::GetEmpty(), in(), CBlast4_subject_Base::IsDatabase(), CBlast4_queries_Base::IsPssm(), NCBI_THROW, CRef< C, Locker >::NotEmpty(), prog, s_ImportDatabase(), s_ImportPssm(), s_ImportQueries(), s_ImportSubjects(), sfree, true, and Warning().

Referenced by RecoverSearchStrategy().

s_ImportSubjects()

static CRef<blast::CBlastDatabaseArgs> s_ImportSubjects ( const CBlast4_subject &  subj, bool  subject_is_protein  )

static

Import the subject sequences into a CBlastDatabaseArgs object.

Definition at line 519 of file blast_app_util.cpp.

References _ASSERT, CBlast4_subject_Base::GetSequences(), CBlast4_subject_Base::IsSequences(), s_ConvertBioseqs2TSeqLocVector(), and CBlastDatabaseArgs::SetSubjects().

Referenced by s_ImportSearchStrategy().

s_InitializeExportStrategy()

static CRef<blast::CExportStrategy> s_InitializeExportStrategy ( CRef< blast::IQueryFactory >  queries, CRef< blast::CBlastDatabaseArgs >  db_args, CRef< blast::CBlastOptionsHandle >  opts_hndl, const string &  client_id, CRef< objects::CPssmWithParameters >  pssm, unsigned int  num_iters  )

static

Definition at line 264 of file blast_app_util.cpp.

References _ASSERT, CRef< C, Locker >::Empty(), NCBI_THROW, CRef< C, Locker >::NotEmpty(), and CRef< C, Locker >::Reset().

Referenced by s_ExportSearchStrategy().

s_IsPrefetchFormat()

static bool s_IsPrefetchFormat ( blast::CFormattingArgs::EOutputFormat  format_type )

static

Definition at line 716 of file blast_app_util.cpp.

References CFormattingArgs::eArchiveFormat, CFormattingArgs::eAsnBinary, CFormattingArgs::eAsnText, and CFormattingArgs::eJsonSeqalign.

Referenced by s_PreFetchSeqs().

s_IsUsingRemoteBlastDbDataLoader()

static bool s_IsUsingRemoteBlastDbDataLoader ( )

static

Returns true if the remote BLAST DB data loader is being used.

Definition at line 703 of file blast_app_util.cpp.

References CObjectManager::GetInstance(), CObjectManager::GetRegisteredNames(), ITERATE, and NStr::StartsWith().

Referenced by BlastFormatter_PreFetchSequenceData().

s_PreFetchSeqs()

static bool s_PreFetchSeqs ( const blast::CSearchResultSet &  results, blast::CFormattingArgs::EOutputFormat  format_type  )

static

Definition at line 728 of file blast_app_util.cpp.

References i, results, s_IsPrefetchFormat(), and NStr::StringToInt().

Referenced by BlastFormatter_PreFetchSequenceData().

SaveSearchStrategy()

void SaveSearchStrategy	(	const CArgs &	args,
		blast::CBlastAppArgs *	cmdline_args,
		CRef< blast::IQueryFactory >	queries,
		CRef< blast::CBlastOptionsHandle >	opts_hndl,
		CRef< objects::CPssmWithParameters >	pssm,
		unsigned int	num_iters
	)

Save the search strategy corresponding to the current command line search.

Definition at line 636 of file blast_app_util.cpp.

References out(), and s_ExportSearchStrategy().

Referenced by CPsiBlastApp::DoIterations(), CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CDeltaBlastApp::Run(), CTblastxApp::Run(), CRMBlastnApp::Run(), CVDBBlastnApp::Run(), CVDBTblastnApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

UseXInclude()

bool UseXInclude	(	const CFormattingArgs &	f,
		const string &	s
	)

Definition at line 887 of file blast_app_util.cpp.

References CFormattingArgs::eJson, CFormattingArgs::eXml2, f, and NCBI_THROW.

Referenced by CBlastnNode::Main(), CBlastpNode::Main(), CBlastxNode::Main(), CRPSBlastNode::Main(), CRPSTBlastnNode::Main(), CTblastnNode::Main(), CBlastFormatterApp::PrintFormattedOutput(), CBlastFormatterVdbApp::PrintFormattedOutput(), CDeltaBlastApp::Run(), CPsiBlastApp::Run(), CTblastxApp::Run(), CVDBBlastnApp::Run(), CBlastnApp::x_RunMTBySplitDB(), CBlastpApp::x_RunMTBySplitDB(), CBlastxApp::x_RunMTBySplitDB(), CRPSBlastApp::x_RunMTBySplitDB(), CRPSTBlastnApp::x_RunMTBySplitDB(), and CTblastnApp::x_RunMTBySplitDB().

USING_SCOPE() [1/2]

USING_SCOPE

(

blast

)

USING_SCOPE() [2/2]

USING_SCOPE

(

objects

)