makeblastdb.cpp File Reference

Command line tool to create BLAST databases. More...

#include <ncbi_pch.hpp>
#include <serial/objostrjson.hpp>
#include <algo/blast/api/version.hpp>
#include <algo/blast/blastinput/blast_input_aux.hpp>
#include <corelib/ncbiapp.hpp>
#include <serial/iterator.hpp>
#include <objmgr/util/create_defline.hpp>
#include <objmgr/util/sequence.hpp>
#include <objects/seq/Seqdesc.hpp>
#include <objects/general/User_object.hpp>
#include <objects/seqfeat/Org_ref.hpp>
#include <objects/blastdb/Blast_db_mask_info.hpp>
#include <objects/blastdb/Blast_mask_list.hpp>
#include <objtools/blast/seqdb_reader/seqdb.hpp>
#include <objtools/blast/seqdb_reader/seqdbcommon.hpp>
#include <objtools/blast/seqdb_writer/writedb.hpp>
#include <objtools/blast/seqdb_writer/writedb_error.hpp>
#include <util/format_guess.hpp>
#include <util/util_exception.hpp>
#include <util/compress/stream_util.hpp>
#include <objtools/blast/seqdb_writer/build_db.hpp>
#include <algo/blast/blastinput/blast_input.hpp>
#include "../blast/blast_app_util.hpp"
#include "masked_range_set.hpp"
#include <iostream>
#include <sstream>
#include <fstream>
#include <random>

Include dependency graph for makeblastdb.cpp:

Go to the source code of this file.

Go to the SVN repository for this file.

Classes
class	CMakeBlastDBApp
	The main application class. More...
class	CSeqEntrySource
class	CRawSeqDBSource

Functions
	USING_SCOPE (blast)
	USING_SCOPE (objects)
template<class TObj >
void	s_ReadObject (CNcbiIstream &file, CFormatGuess::EFormat fmt, CRef< TObj > &obj, const string &msg)
	Reads an object defined in a NCBI ASN.1 spec from a stream in multiple formats: binary and text ASN.1 and XML. More...
template<class TObj >
void	s_ReadObject (CNcbiIstream &file, CRef< TObj > &obj, const string &msg)
	Overloaded version of s_ReadObject which uses CFormatGuess to determine the encoding of the object in the file. More...
static const string	kInput ("in")
	Command line flag to represent the input. More...
static const string	kInputSeparators (" ")
	Defines token separators when multiple inputs are present. More...
static const string	kOutput ("out")
	Command line flag to represent the output. More...
static string	Uint8ToString_DataSize (Uint8 v, unsigned minprec=10)
	Converts a Uint8 into a string which contains a data size (converse to NStr::StringToUInt8_DataSize) More...
static TTaxId	s_GetTaxId (const CBioseq &bio)
static bool	s_HasTitle (const CBioseq &bio)
int	main (int argc, const char *argv[])

Variables
	USING_NCBI_SCOPE

Detailed Description

Command line tool to create BLAST databases.

This is the successor to formatdb from the C toolkit

Definition in file makeblastdb.cpp.

Function Documentation

kInput()

static const string kInput ( "in"  )

static

Command line flag to represent the input.

Referenced by CMakeBlastDBApp::Init(), and CMakeBlastDBApp::x_BuildDatabase().

kInputSeparators()

static const string kInputSeparators ( " "  )

static

Defines token separators when multiple inputs are present.

Referenced by CMakeBlastDBApp::x_BuildDatabase().

kOutput()

static const string kOutput ( "out"  )

static

Command line flag to represent the output.

Referenced by CMakeBlastDBApp::Init(), and CMakeBlastDBApp::x_BuildDatabase().

main()

int main	(	int	argc,
		const char *	argv[]
	)

Definition at line 1357 of file makeblastdb.cpp.

References CNcbiApplicationAPI::AppMain().

s_GetTaxId()

static TTaxId s_GetTaxId ( const CBioseq &  bio )

static

Definition at line 367 of file makeblastdb.cpp.

References CSeq_descr_Base::Get(), CSeq_entry::GetDescr(), CSeqdesc_Base::GetOrg(), CBioSource_Base::GetOrg(), CBioseq::GetParentEntry(), CSeq_entry::GetParentEntry(), CSeqdesc_Base::GetSource(), COrg_ref::GetTaxId(), CSeqdesc_Base::IsOrg(), CSeq_entry::IsSetDescr(), CSeqdesc_Base::IsSource(), ITERATE, NULL, and ZERO_TAX_ID.

Referenced by CSeqEntrySource::GetNext().

s_HasTitle()

static bool s_HasTitle ( const CBioseq &  bio )

static

Definition at line 391 of file makeblastdb.cpp.

References CBioseq_Base::CanGetDescr(), CSeq_descr_Base::Get(), CBioseq_Base::GetDescr(), CSeqdesc_Base::IsTitle(), and ITERATE.

Referenced by CSeqEntrySource::GetNext().

s_ReadObject() [1/2]

template<class TObj >

void s_ReadObject	(	CNcbiIstream &	file,
		CFormatGuess::EFormat	fmt,
		CRef< TObj > &	obj,
		const string &	msg
	)

Reads an object defined in a NCBI ASN.1 spec from a stream in multiple formats: binary and text ASN.1 and XML.

Parameters

file	stream to read the object from [in]
fmt	specifies the format in which the object is encoded [in]
obj	on input is an empty CRef<> object, on output it's populated with the object read [in|out]
msg	error message to display if reading fails [in]

Definition at line 164 of file makeblastdb.cpp.

References CFormatGuess::eBinaryASN, CFormatGuess::eTextASN, file, MSerial_AsnBinary, MSerial_AsnText, msg(), NCBI_THROW, and CRef< C, Locker >::Reset().

Referenced by CSeqEntrySource::CSeqEntrySource(), s_ReadObject(), and CMakeBlastDBApp::x_ProcessMaskData().

s_ReadObject() [2/2]

template<class TObj >

void s_ReadObject	(	CNcbiIstream &	file,
		CRef< TObj > &	obj,
		const string &	msg
	)

Overloaded version of s_ReadObject which uses CFormatGuess to determine the encoding of the object in the file.

Parameters

file	stream to read the object from [in]
obj	on input is an empty CRef<> object, on output it's populated with the object read [in|out]
msg	error message to display if reading fails [in]

Definition at line 192 of file makeblastdb.cpp.

References CFormatGuess::eBinaryASN, CFormatGuess::eTextASN, file, msg(), and s_ReadObject().

Uint8ToString_DataSize()

static string Uint8ToString_DataSize ( Uint8  v, unsigned  minprec = 10  )

static

Converts a Uint8 into a string which contains a data size (converse to NStr::StringToUInt8_DataSize)

Parameters

v	value to convert [in]
minprec	minimum precision [in]

Definition at line 341 of file makeblastdb.cpp.

References i, and NStr::UInt8ToString().

Referenced by CMakeBlastDBApp::x_BuildDatabase().

USING_SCOPE() [1/2]

USING_SCOPE

(

blast

)

USING_SCOPE() [2/2]

USING_SCOPE

(

objects

)

Variable Documentation

USING_NCBI_SCOPE

USING_NCBI_SCOPE

Definition at line 71 of file makeblastdb.cpp.