Representation of a microarray dataset that is used by Seek. More...

#include <seekdataset.h>

Public Types
enum	DistanceMeasure { CORRELATION = 0, Z_SCORE = CORRELATION + 1 }
	Distance measure (see main section for descriptions) More...
Public Member Functions
	CSeekDataset ()
	Constructor.
	~CSeekDataset ()
	Destructor.
bool	ReadDatasetAverageStdev (const string &)
	Read the `*`.sinfo file.
bool	ReadGeneAverage (const string &)
	Read the gene average correlation file `*`.gavg.
bool	ReadGeneVariance (const string &)
	Read the gene variance file `*`.gvar.
bool	ReadGenePresence (const string &)
	Read the gene presence file `*`.gpres.
bool	InitializeGeneMap ()
	Initialize the genome presence map.
bool	InitializeQuery (const vector< utype > &)
	Initialize the query presence map.
bool	InitializeQueryBlock (const vector< utype > &)
	Initialize a presence map for a block of queries.
bool	DeleteQuery ()
	Delete the query.
bool	DeleteQueryBlock ()
	Delete query block.
bool	InitializeDataMatrix (utype *, const vector< float > &, const utype &, const utype &, const bool=true, const bool=false, const bool=false, const enum DistanceMeasure=Z_SCORE, const float cutoff=-1.0 CMeta::GetNaN(), const bool=false, gsl_rng *rand=NULL)
	Initialize the gene-gene correlation matrix.
bool	Copy (CSeekDataset *)
	Copy constructor.
utype **	GetDataMatrix ()
	Get the gene-gene correlation matrix.
unsigned char **	GetMatrix ()
	Get the gene-gene correlation matrix.
CSeekIntIntMap *	GetGeneMap ()
	Get the genome presence map.
CSeekIntIntMap *	GetDBMap ()
	Get the query-block presence map.
CSeekIntIntMap *	GetQueryMap ()
	Get the query presence map.
const vector< utype > &	GetQuery () const
	Get the query genes.
const vector< utype > &	GetQueryIndex () const
	Get the query gene indices.
float	GetGeneVariance (const utype &) const
	Get the gene expression variance vector.
float	GetGeneAverage (const utype &) const
	Get the gene average correlation vector.
float	GetDatasetAverage () const
	Get the mean of the global gene-gene Pearson distribution.
float	GetDatasetStdev () const
	Get the standard deviation of the global gene-gene Pearson distribution.
utype	GetNumGenes () const
	Get the genome size.
bool	InitializeCVWeight (const utype &)
	Initialize the weight of the dataset.
bool	SetCVWeight (const utype &, const float &)
	Set the score for a particular cross-validation.
float	GetCVWeight (const utype &)
	Get the score for a particular cross-validation.
const vector< float > &	GetCVWeight () const
	Get all the cross-validation scores.
float	GetDatasetSumWeight ()
	Get the dataset weight.
void	SetPlatform (CSeekPlatform &)
	Set the platform.
CSeekPlatform &	GetPlatform () const
	Get the platform.

Detailed Description

Representation of a microarray dataset that is used by Seek.

A CSeekDataset encapsulates the following information about the dataset:

The gene-gene correlation matrix
Each gene's expression variance
Each gene's average correlation
The genes in the query that are present in the dataset
The genes in the genome that are present in the dataset
The platform which the dataset belongs to
The weight of the dataset that is assigned by the search algorithm

This dataset structure is designed to be used by Seek.

Remarks:

The word correlation refers to the standardized z-scores of Pearson correlations, which is derived from a 2-step process:

$f(x,y)=\frac{1}{2}ln\frac{1+p(x,y)}{1-p(x,y)}$

where

is the Pearson correlation, $f(x,y)$

is the Fisher's transformed score.

$z(x,y)=\frac{f(x,y) - \bar{f}}{\sigma_{f}}$

where

is the z-score, $\bar{f}$ is the mean, and $\sigma_{f}$ is the standard deviation.

From here on, correlation always refers to the above z-score definition.

Definition at line 133 of file seekdataset.h.

Member Enumeration Documentation

enum Sleipnir::CSeekDataset::DistanceMeasure

Distance measure (see main section for descriptions)

Enumerator:

CORRELATION	Pearson correlations
Z_SCORE	Z-score of Pearson correlations

Definition at line 140 of file seekdataset.h.

Member Function Documentation

bool Sleipnir::CSeekDataset::Copy ( CSeekDataset * src )

Copy constructor.

Parameters:

src	A given dataset

Definition at line 67 of file seekdataset.cpp.

bool Sleipnir::CSeekDataset::DeleteQuery ( )

Delete the query.

Resets all query-related data, such as dataset weight, query presence map, etc.

Definition at line 240 of file seekdataset.cpp.

Referenced by DeleteQueryBlock(), InitializeQuery(), and ~CSeekDataset().

bool Sleipnir::CSeekDataset::DeleteQueryBlock ( )

Delete query block.

Resets all query-block related data.

Definition at line 255 of file seekdataset.cpp.

References DeleteQuery(), and Sleipnir::CSeekTools::Free2DArray().

Referenced by InitializeQueryBlock(), and ~CSeekDataset().

float Sleipnir::CSeekDataset::GetCVWeight ( const utype & i )

Get the score for a particular cross-validation.

Parameters:

i The index

Definition at line 572 of file seekdataset.cpp.

const vector< float > & Sleipnir::CSeekDataset::GetCVWeight ( ) const

Get all the cross-validation scores.

Returns:: A vector of cross-validation scores

Definition at line 576 of file seekdataset.cpp.

utype ** Sleipnir::CSeekDataset::GetDataMatrix ( )

Get the gene-gene correlation matrix.

Returns:: A two-dimensional array of type utype. Note that the correlation has been scaled to a integer range from 0 to 640. See CSeekDataset::InitializeDataMatrix.

Definition at line 277 of file seekdataset.cpp.

Referenced by Sleipnir::CSeekWeighter::LinearCombine().

float Sleipnir::CSeekDataset::GetDatasetAverage ( ) const

Get the mean of the global gene-gene Pearson distribution.

Returns:: The mean Pearson for the dataset

Definition at line 540 of file seekdataset.cpp.

float Sleipnir::CSeekDataset::GetDatasetStdev ( ) const

Get the standard deviation of the global gene-gene Pearson distribution.

Returns:: The standard deviation of the Pearson distribution

Definition at line 544 of file seekdataset.cpp.

float Sleipnir::CSeekDataset::GetDatasetSumWeight ( )

Get the dataset weight.

Returns:: The dataset weight

Definition at line 580 of file seekdataset.cpp.

CSeekIntIntMap * Sleipnir::CSeekDataset::GetDBMap ( )

Get the query-block presence map.

Returns:: The query-block presence map

Definition at line 536 of file seekdataset.cpp.

float Sleipnir::CSeekDataset::GetGeneAverage ( const utype & i ) const

Get the gene average correlation vector.

Returns:: The average correlation vector

Definition at line 552 of file seekdataset.cpp.

Referenced by InitializeDataMatrix().

CSeekIntIntMap * Sleipnir::CSeekDataset::GetGeneMap ( )

Get the genome presence map.

Returns:: The genome presence map

Definition at line 528 of file seekdataset.cpp.

Referenced by Sleipnir::CSeekWeighter::CVWeighting(), Sleipnir::CSeekWeighter::LinearCombine(), and Sleipnir::CSeekWeighter::OneGeneWeighting().

float Sleipnir::CSeekDataset::GetGeneVariance ( const utype & i ) const

Get the gene expression variance vector.

Returns:: The variance vector

Definition at line 548 of file seekdataset.cpp.

unsigned char ** Sleipnir::CSeekDataset::GetMatrix ( )

Get the gene-gene correlation matrix.

Returns:: A two-dimensional array of type unsigned char**.

Definition at line 524 of file seekdataset.cpp.

utype Sleipnir::CSeekDataset::GetNumGenes ( ) const

Get the genome size.

Returns:: The genome size

Definition at line 556 of file seekdataset.cpp.

Referenced by Sleipnir::CSeekWeighter::CVWeighting(), Sleipnir::CSeekWeighter::LinearCombine(), and Sleipnir::CSeekWeighter::OneGeneWeighting().

CSeekPlatform & Sleipnir::CSeekDataset::GetPlatform ( ) const

Get the platform.

Returns:: The platform of this dataset

Definition at line 599 of file seekdataset.cpp.

const vector< utype > & Sleipnir::CSeekDataset::GetQuery ( ) const

Get the query genes.

Returns:: A vector of queries

Definition at line 269 of file seekdataset.cpp.

const vector< utype > & Sleipnir::CSeekDataset::GetQueryIndex ( ) const

Get the query gene indices.

Returns:: A vector of query gene indices

Definition at line 273 of file seekdataset.cpp.

CSeekIntIntMap * Sleipnir::CSeekDataset::GetQueryMap ( )

Get the query presence map.

Returns:: The query presence map

Definition at line 532 of file seekdataset.cpp.

Referenced by Sleipnir::CSeekWeighter::CVWeighting(), Sleipnir::CSeekWeighter::LinearCombine(), and Sleipnir::CSeekWeighter::OneGeneWeighting().

bool Sleipnir::CSeekDataset::InitializeCVWeight ( const utype & i )

Initialize the weight of the dataset.

Parameters:

i	The number of cross-validations

Initializes the total dataset weight, and the score of the individual cross-validation (CV) runs.

Definition at line 560 of file seekdataset.cpp.

Referenced by Sleipnir::CSeekWeighter::CVWeighting(), and Sleipnir::CSeekWeighter::OneGeneWeighting().

bool Sleipnir::CSeekDataset::InitializeDataMatrix	(	utype **	,
		const vector< float > &	,
		const utype &	,
		const utype &	,
		const bool	= `true`,
		const bool	= `false`,
		const bool	= `false`,
		const enum	DistanceMeasure = `Z_SCORE`,
		const float	cutoff = `-1.0*CMeta::GetNaN()`,
		const bool	= `false`,
		gsl_rng *	rand = `NULL`
	)

Initialize the gene-gene correlation matrix.

Parameters:

rD	A two-dimensional array storing the discretized gene-gene correlations
quant	The discretization function
iRows	The number of rows for the correlation matrix
iColumns	The number of columns for the correlation matrix
bSubtractAvg	If true, subtract the correlation by the dataset average
bNormPlatform	If true, subtract the correlation by the platform average and divide by standard deviation
logit	If true, apply the logit transform on correlations
dist_measure	Distance measure: z-score or correlations
cutoff	Apply a hard cutoff on correlations
bRandom	If true, shuffle the correlation vector
rand	The random generator for the shuffling operation above

Remarks:

The discretized correlation in the matrix rD is bounded by 0 to 255 (the limit of unsigned char). The parameter quant specifies how a correlation is discretized. For example, if the quant has 5 bins: