A suite of file I/O and general purpose tools that are used by Seek. More...

#include <seekreader.h>

Static Public Member Functions
template<class tType >
static bool	ReadArray (const char *fileName, vector< tType > &vData)
	Read an array from a given binary file.
template<class tType >
static bool	WriteArray (const char *fileName, const vector< tType > &vData)
	Write an array in binary format.
template<class tType >
static bool	WriteArrayText (const char *fileName, const vector< tType > &vData)
	Write an array in text format.
template<class tType >
static bool	Write2DArrayText (const char *fileName, const vector< vector< tType > > &vData)
	Write a two-dimensional array in text format.
template<class tType >
static bool	InitVector (vector< tType > &vData, const utype &iSize, const tType &tValue)
	Initialize a vector with a given value.
template<class tType >
static bool	InitVector (vector< tType > &vData, const utype &iSize)
	Initialize a vector.
template<class tType >
static tType **	Init2DArray (const size_t &iSize1, const size_t &iSize2, const tType &tValue)
	Initialize a two-dimensional array with the given size and value.
template<class tType >
static void	Free2DArray (tType **f)
	Free a two-dimensional array.
static bool	IsNaN (const utype &)
	Checks if a `utype` value is invalid.
static utype	GetNaN ()
	Return the NaN value as a utype.
static string	ConvertInt (const int &)
	Converts an integer to a string.
static bool	ReadDatabaselets (const vector< CDatabase * > &, const size_t &, const size_t &, const vector< vector< string > > &, vector< CSeekDataset * > &, const map< string, utype > &, const vector< vector< string > > &, const map< string, utype > &, const int &, const bool &)
	Read a set of CDatabaselet from CDatabase instance.
static bool	LoadDatabase (const vector< CDatabase * > &, const size_t &, const size_t &, const vector< CSeekDBSetting * > &, const vector< string > &, const map< string, string > &, const map< string, utype > &, vector< CSeekPlatform > &, vector< CSeekDataset * > &, const vector< vector< string > > &, const map< string, utype > &, const bool=false, const bool=false)
	Read the search setting files and load the CDatabase.
static bool	LoadDatabase (const vector< CDatabase * > &, const size_t &, const size_t &, vector< CSeekDataset * > &, const vector< CSeekDataset * > &, vector< CSeekPlatform > &, const vector< CSeekPlatform > &, const vector< string > &, const map< string, string > &, const map< string, utype > &)
	Load a CDatabase by copying from an existing instance.
static bool	ReadPlatforms (const string &strPlatformDirectory, vector< CSeekPlatform > &plat, vector< string > &vecstrPlatforms, map< string, utype > &mapstriPlatforms, const int lineSize=1024)
	Read the platforms.
static bool	ReadPlatforms (const char *plat_dir, vector< CSeekPlatform > &plat, vector< string > &vecstrPlatforms, map< string, utype > &mapstriPlatforms, const int lineSize=1024)
	Read the platforms.
static bool	ReadListOneColumn (const string &strFile, vector< string > &vecstrList, CSeekStrIntMap &mapstriList, const int lineSize=1024)
	Read a table with one column.
static bool	ReadListOneColumn (const char *file, vector< string > &vecstrList, CSeekStrIntMap &mapstriList, const int lineSize=1024)
	Read a table with one column.
static bool	ReadListOneColumn (const string &strFile, vector< string > &vecstrList, const int lineSize=1024)
	Read a table with one column.
static bool	ReadListOneColumn (const char *file, vector< string > &vecstrList, const int lineSize=1024)
	Read a table with one column.
static bool	ReadListTwoColumns (const string &strFile, vector< string > &list1, vector< string > &list2, const int lineSize=1024)
	Read a table with two columns.
static bool	ReadListTwoColumns (const char *file, vector< string > &list1, vector< string > &list2, const int lineSize=1024)
	Read a table with two columns.
static bool	ReadMultipleQueries (const string &strFile, vector< vector< string > > &qList, const int lineSize=1024)
	Read a list of queries.
static bool	ReadMultipleNotQueries (const char *file, vector< vector< vector< string > > > &qList, const int lineSize=1024)
static bool	ReadMultipleQueries (const char *file, vector< vector< string > > &qList, const int lineSize=1024)
	Read a list of queries.
static bool	ReadMultiGeneOneLine (const string &strFile, vector< string > &list1, const int lineSize=1024)
	Read just one gene-set line.
static bool	ReadMultiGeneOneLine (const char *file, vector< string > &list1, const int lineSize=1024)
	Read just one gene-set line.
static bool	ReadQuantFile (const string &strFile, vector< float > &quant, const int lineSize=5000)
	Read the correlation discretization.
static bool	ReadQuantFile (const char *file, vector< float > &quant, const int lineSize=5000)
	Read the correlation discretization.

Detailed Description

A suite of file I/O and general purpose tools that are used by Seek.

These tools are critical for initializing the search parameters, and are highly beneficial to the routine manipulations of vectors and files.

Some examples of these tools include:

Reading different search setting files, such as the dataset mapping, the gene mapping, the list of queries, the correlation discretization file, etc.
Loading a set of CDatabaselet from the given directory
Reading a binary file that contains a vector of standard type elements (int, char, string, float)
Reading a text file that contains a table with one or two columns
Setting all elements of a given vector to the given value
Writing a vector to a file (in binary or in text format)

Definition at line 52 of file seekreader.h.

Member Function Documentation

string Sleipnir::CSeekTools::ConvertInt ( const int & number ) [static]

Converts an integer to a string.

Parameters:

number The given integer number

Returns:: The string

Definition at line 26 of file seekreader.cpp.

Referenced by ReadDatabaselets().

template<class tType >

static void Sleipnir::CSeekTools::Free2DArray ( tType ** f ) [inline, static]

Free a two-dimensional array.

Parameters:

f	The two-dimensional array

Definition at line 255 of file seekreader.h.

Referenced by Sleipnir::CSeekDataset::DeleteQueryBlock(), Sleipnir::CSeekDataset::InitializeDataMatrix(), Sleipnir::CSeekWeighter::OrderStatisticsRankAggregation(), and Sleipnir::CSeekCentral::~CSeekCentral().

template<class tType >

static tType** Sleipnir::CSeekTools::Init2DArray	(	const size_t &	iSize1,
		const size_t &	iSize2,
		const tType &	tValue
	)		`[inline, static]`

Initialize a two-dimensional array with the given size and value.

Parameters:

iSize1	The first dimension size
iSize2	The second dimension size
tValue	The value

Creates a two-dimensional array of the given dimension, then populates it with the given value.

Definition at line 224 of file seekreader.h.

Referenced by Sleipnir::CSeekDataset::InitializeDataMatrix(), Sleipnir::CSeekDataset::InitializeQueryBlock(), and Sleipnir::CSeekWeighter::OrderStatisticsRankAggregation().

template<class tType >

static bool Sleipnir::CSeekTools::InitVector	(	vector< tType > &	vData,
		const utype &	iSize,
		const tType &	tValue
	)		`[inline, static]`

Initialize a vector with a given value.

Parameters:

vData	The source vector
iSize	The number of elements that the vector should contain
tValue	The value

Resizes the source vector to the given size, then sets all elements in the vector to the given value.

Definition at line 190 of file seekreader.h.

Referenced by Sleipnir::CSeekWeighter::CVWeighting(), Sleipnir::CSeekDataset::InitializeDataMatrix(), Sleipnir::CSeekPlatform::InitializePlatform(), Sleipnir::CSeekWeighter::LinearCombine(), Sleipnir::CSeekWeighter::OneGeneWeighting(), ReadDatabaselets(), and Sleipnir::CSeekCentral::VarianceWeightSearch().

template<class tType >

static bool Sleipnir::CSeekTools::InitVector	(	vector< tType > &	vData,
		const utype &	iSize
	)		`[inline, static]`

Initialize a vector.

Parameters:

vData	The source vector
iSize	The number of elements that the vector should contain

Resizes the source vector to the given size.

Definition at line 207 of file seekreader.h.

bool Sleipnir::CSeekTools::IsNaN ( const utype & v ) [static]

Checks if a utype value is invalid.

Parameters:

v	The value to be checked A `utype` value is invalid if it is maximum (65535).

Definition at line 32 of file seekreader.cpp.

Referenced by Sleipnir::CSeekQuery::CreateCVPartitions(), Sleipnir::CSeekWeighter::CVWeighting(), Sleipnir::CSeekDataset::InitializeQuery(), Sleipnir::CSeekDataset::InitializeQueryBlock(), Sleipnir::CSeekWeighter::OneGeneWeighting(), and ReadDatabaselets().

bool Sleipnir::CSeekTools::LoadDatabase	(	const vector< CDatabase * > &	DB,
		const size_t &	iGenes,
		const size_t &	iDatasets,
		const vector< CSeekDBSetting * > &	DBSetting,
		const vector< string > &	vecstrDatasets,
		const map< string, string > &	mapstrstrDatasetPlatform,
		const map< string, utype > &	mapstriPlatform,
		vector< CSeekPlatform > &	vp,
		vector< CSeekDataset * > &	vc,
		const vector< vector< string > > &	dbDataset,
		const map< string, utype > &	mapstriDataset,
		const bool	bVariance = `false`,
		const bool	bCorrelation = `false`
	)		`[static]`

Read the search setting files and load the CDatabase.

Performs the following search initializing operations:

Reads the gene presence files *.gpres, the gene averages *.gavg, the gene variances *.gvar, and each dataset's correlation average and standard deviation *.sinfo.
Reads the dataset-platform mapping file
Initializes the vector of CSeekDataset with each dataset's gene-presence, gene-averages

Parameters:

DB	The CDatabase instance
strPrepInputDirectory	The prep directory which contains the ``.gavg and ``.gpres files
strGvarInputDirectory	The directory that contains the gene variance files `*`.gvar
strSinfoInputDirectory	The directory that contains the `*`.sinfo files
vecstrDatasets	The dataset definition
mapstrstrDatasetPlatform	The dataset-platform mapping
mapstriPlatform	Platform name-platform ID mapping
vp	The vector of CSeekPlatform
vc	The vector of CSeekDataset, the output

Definition at line 215 of file seekreader.cpp.

Referenced by Sleipnir::CSeekCentral::Initialize().

bool Sleipnir::CSeekTools::LoadDatabase	(	const vector< CDatabase * > &	DB,
		const size_t &	iGenes,
		const size_t &	iDatasets,
		vector< CSeekDataset * > &	vc,
		const vector< CSeekDataset * > &	vc_src,
		vector< CSeekPlatform > &	vp,
		const vector< CSeekPlatform > &	vp_src,
		const vector< string > &	vecstrDatasets,
		const map< string, string > &	mapstrstrDatasetPlatform,
		const map< string, utype > &	mapstriPlatform
	)		`[static]`

Load a CDatabase by copying from an existing instance.

Copies the vector of initialized CSeekDataset to a new vector. Copies the vector of initialized CSeekPlatform to a new vector.

Parameters:

DB	The CDatabase
vc	The destination dataset vector
vc_src	The source dataset vector
vp	The destination platform vector
vp_src	The source platform vector
vecstrDatasets	The dataset definition
mapstrstrDatasetPlatform	The dataset-platform mapping
mapstriPlatform	Platform name-platform ID mapping

Definition at line 176 of file seekreader.cpp.

template<class tType >

static bool Sleipnir::CSeekTools::ReadArray	(	const char *	fileName,
		vector< tType > &	vData
	)		`[inline, static]`

Read an array from a given binary file.

Parameters:

fileName	The file name
vData	The destination array

Returns:: True if the reading is successful

Remarks:: This function reads an one-dimensional array. The binary file needs to be organized as follows: 1) The first field is the size of the array, N (size_t). 2) The second field is a set of N elements.

Definition at line 67 of file seekreader.h.

Referenced by Sleipnir::CSeekDataset::ReadDatasetAverageStdev(), Sleipnir::CSeekDataset::ReadGeneAverage(), Sleipnir::CSeekDataset::ReadGenePresence(), and Sleipnir::CSeekDataset::ReadGeneVariance().

bool Sleipnir::CSeekTools::ReadDatabaselets	(	const vector< CDatabase * > &	DB,
		const size_t &	iGenes,
		const size_t &	iDatasets,
		const vector< vector< string > > &	vecstrAllQuery,
		vector< CSeekDataset * > &	vc,
		const map< string, utype > &	mapstriGenes,
		const vector< vector< string > > &	dbDatasets,
		const map< string, utype > &	mapstriDatasets,
		const int &	iClient,
		const bool &	bNetwork
	)		`[static]`

Read a set of CDatabaselet from CDatabase instance.

Parameters:

DB	The CDatabase instance
vecstrAllQuery	The list of queries
vc	A vector of datasets (the output)
iClient	If the network mode is enabled, the client's socket
bNetwork	If true, the network mode is enabled

Remarks:: A CDatabaselet stores the correlation of a given gene, g, to all other genes in all of the datasets. In order to perform the coexpression search, the CDatabaselet's corresponding to the query genes need to be loaded from disk. This function reads the CDatabaselet files corresponding to the query genes.; Once the CDatabaselet for a query gene has been read, the next step that this function performs is allotting the correlations to their corresponding datasets (CSeekDataset).; The network mode is used to relay status messages between the server and the client.; Assumes that the CSeekTools::LoadDatabase() has been called.

Definition at line 41 of file seekreader.cpp.

References ConvertInt(), Sleipnir::CSeekIntIntMap::GetForward(), InitVector(), IsNaN(), and Sleipnir::CSeekNetwork::Send().

bool Sleipnir::CSeekTools::ReadListOneColumn	(	const string &	strFile,
		vector< string > &	vecstrList,
		CSeekStrIntMap &	mapstriList,
		const int	lineSize = `1024`
	)		`[static]`

Read a table with one column.

Outputs the lines in the table as a vector of strings

Parameters:

strFile	The file name
vecstrList	The output
mapstriList	Mapping the line to its line number
lineSize	The maximum characters per line in the file (default 1024)

Definition at line 387 of file seekreader.cpp.

Referenced by ReadListOneColumn().

bool Sleipnir::CSeekTools::ReadListOneColumn	(	const char *	file,
		vector< string > &	vecstrList,
		CSeekStrIntMap &	mapstriList,
		const int	lineSize = `1024`
	)		`[static]`

Read a table with one column.

This is the same as the previous CSeekTools::ReadListOneColumn() declaration, except that the accepted string arguments are of the type const char*.

Definition at line 394 of file seekreader.cpp.

References Sleipnir::CSeekStrIntMap::Set().

bool Sleipnir::CSeekTools::ReadListOneColumn	(	const string &	strFile,
		vector< string > &	vecstrList,
		const int	lineSize = `1024`
	)		`[static]`

Read a table with one column.

Same as the previous CSeekTools::ReadListOneColumn() declaration, except that this does not generate the line to line number mapping.

Definition at line 541 of file seekreader.cpp.

References ReadListOneColumn().

bool Sleipnir::CSeekTools::ReadListOneColumn	(	const char *	file,
		vector< string > &	vecstrList,
		const int	lineSize = `1024`
	)		`[static]`

Read a table with one column.

Same as the previous CSeekTools::ReadListOneColumn() declaration, except that this does not generate the line to line number mapping, and accepts the file name as const char *.

Definition at line 546 of file seekreader.cpp.

bool Sleipnir::CSeekTools::ReadListTwoColumns	(	const string &	strFile,
		vector< string > &	list1,
		vector< string > &	list2,
		const int	lineSize = `1024`
	)		`[static]`

Read a table with two columns.

Parameters:

strFile	The file name
list1	The column 1 output
list2	The column 2 output
lineSize	The maximum characters per line in the file (default 1024)

Definition at line 352 of file seekreader.cpp.

Referenced by Sleipnir::CSeekCentral::Initialize().

bool Sleipnir::CSeekTools::ReadListTwoColumns	(	const char *	file,
		vector< string > &	list1,
		vector< string > &	list2,
		const int	lineSize = `1024`
	)		`[static]`

Read a table with two columns.

This is the same as the previous CSeekTools::ReadListTwoColumns() declaration, except that the accepted string arguments are of the type const char *.

Definition at line 358 of file seekreader.cpp.

References Sleipnir::CMeta::Tokenize().

bool Sleipnir::CSeekTools::ReadMultiGeneOneLine	(	const string &	strFile,
		vector< string > &	list1,
		const int	lineSize = `1024`
	)		`[static]`

Read just one gene-set line.

Reads the first line in the file. The line contains a set of gene names delimited by spaces. The output is a vector of strings representing the genes in that line.

Parameters:

strFile	The file name
list1	The output
lineSize	The maximum characters per line in the file (default 1024)

Definition at line 508 of file seekreader.cpp.

bool Sleipnir::CSeekTools::ReadMultiGeneOneLine	(	const char *	file,
		vector< string > &	list1,
		const int	lineSize = `1024`
	)		`[static]`

Read just one gene-set line.

Same as the previous CSeekTools::ReadMultiGeneOneLine() except that the accepted string argument is of the type const char *.

Definition at line 513 of file seekreader.cpp.

References Sleipnir::CMeta::Tokenize().

bool Sleipnir::CSeekTools::ReadMultipleQueries	(	const string &	strFile,
		vector< vector< string > > &	qList,
		const int	lineSize = `1024`
	)		`[static]`

Read a list of queries.

A query is specified as a set of gene names delimited by spaces. A query occupies one line in the file.

Parameters:

strFile	The file name
qList	The output
lineSize	The maximum characters per line in the file (default 1024)

Definition at line 422 of file seekreader.cpp.

Referenced by Sleipnir::CSeekCentral::Initialize().

bool Sleipnir::CSeekTools::ReadMultipleQueries	(	const char *	file,
		vector< vector< string > > &	qList,
		const int	lineSize = `1024`
	)		`[static]`

Read a list of queries.

Same as the previous CSeekTools::ReadMultipleQueries() declaration, except that this function accepts the string argument as a const char *.

Definition at line 427 of file seekreader.cpp.

References Sleipnir::CMeta::Tokenize().

bool Sleipnir::CSeekTools::ReadPlatforms	(	const string &	strPlatformDirectory,
		vector< CSeekPlatform > &	plat,
		vector< string > &	vecstrPlatforms,
		map< string, utype > &	mapstriPlatforms,
		const int	lineSize = `1024`
	)		`[static]`

Read the platforms.

Reading the platforms mainly involves reading the correlation average and the correlation standard deviation for each platform in the database. The purpose is to correct the platform specific biases on the correlation values.

Parameters:

strPlatformDirectory	The directory that contains the platform average and standard deviation files
plat	The output
vecstrPlatforms	The platform names
mapstriPlatform	The platform name - platform ID mapping
lineSize	The maximum characters per line in the file (default 1024)

Definition at line 296 of file seekreader.cpp.

Referenced by Sleipnir::CSeekCentral::Initialize().

bool Sleipnir::CSeekTools::ReadPlatforms	(	const char *	plat_dir,
		vector< CSeekPlatform > &	plat,
		vector< string > &	vecstrPlatforms,
		map< string, utype > &	mapstriPlatforms,
		const int	lineSize = `1024`
	)		`[static]`

Read the platforms.

This is the same as the previous CSeekTools::ReadPlatforms() declaration, except that the accepted string arguments are of the type const char *.

Definition at line 303 of file seekreader.cpp.

References Sleipnir::CFullMatrix< tType >::Get(), Sleipnir::CFullMatrix< tType >::GetColumns(), Sleipnir::CFullMatrix< tType >::GetRows(), and Sleipnir::CFullMatrix< tType >::Open().

bool Sleipnir::CSeekTools::ReadQuantFile	(	const string &	strFile,
		vector< float > &	quant,
		const int	lineSize = `5000`
	)		`[static]`

Read the correlation discretization.

Specifies how the correlations should be binned. The file contains the bin boundaries separated by spaces.

Parameters:

strFile	The file name
quant	The output
lineSize	The maximum characters per line in the file (default 5000)

Definition at line 150 of file seekreader.cpp.

Referenced by Sleipnir::CSeekCentral::Initialize().

bool Sleipnir::CSeekTools::ReadQuantFile	(	const char *	file,
		vector< float > &	quant,
		const int	lineSize = `5000`
	)		`[static]`

Read the correlation discretization.

Same as the previous CSeekTools::ReadQuantFile() except that the accepted string argument is of the type const char *.

Definition at line 155 of file seekreader.cpp.

References Sleipnir::CMeta::Tokenize().

template<class tType >

static bool Sleipnir::CSeekTools::Write2DArrayText	(	const char *	fileName,
		const vector< vector< tType > > &	vData
	)		`[inline, static]`

Write a two-dimensional array in text format.

Parameters:

fileName	The file name
vData	The source array

Returns:: True if the reading is successful

Remarks:: This function writes an two-dimensional array in the text format. The rows are separated by new lines. The elements in a row are separated by spaces.

Definition at line 160 of file seekreader.h.

template<class tType >

static bool Sleipnir::CSeekTools::WriteArray	(	const char *	fileName,
		const vector< tType > &	vData
	)		`[inline, static]`

Write an array in binary format.

Parameters:

fileName	The file name
vData	The source array

Returns:: True if the reading is successful

Remarks:: This function writes an one-dimensional array. It will write the array in the following way: 1) The first field is the size of the array, N (size_t). 2) The second field is the N elements in the array.

Definition at line 107 of file seekreader.h.

template<class tType >

static bool Sleipnir::CSeekTools::WriteArrayText	(	const char *	fileName,
		const vector< tType > &	vData
	)		`[inline, static]`

Write an array in text format.

Parameters:

fileName	The file name
vData	The source array

Returns:: True if the reading is successful

Remarks:: This function writes an one-dimensional array in the text format. The array elements are separated by spaces.

Definition at line 137 of file seekreader.h.

The documentation for this class was generated from the following files:

src/seekreader.h
src/seekreader.cpp

Static Public Member Functions

Detailed Description

Member Function Documentation