Biorithm  1.1
Public Member Functions | Friends
MyPointSet Class Reference

Stores a set of points and information about the point's dimensions. This class is typically useful for storing one gene expression dataset. More...

#include <point.h>

List of all members.

Public Member Functions

 MyPointSet (const MyPointSet &copy)
const MyPointSet & operator= (const MyPointSet &rhs)
void compareTTest (string name, const MyPointSet &other, MyNT &tstat, MyNT &pvalue) const
 Compare the expression values of point "name" in the invocant and in other by performing a t-test.
MyNT computeAverageExpressionValue (const set< string > &subset)
void computeCoordinateDistribution (MyHistogram &histogram) const
 Compute the distribution of coordinates in the point set.
MyNT computeMean (string point) const
 Compute the mean value of the vectors corresponding to point.
bool computeMean (string point, MyPoint &mean) const
 Compute the mean (vector) of the vectors corresponding to point.
bool computeMeanAndVariance (string point, MyNT &mu, MyNT &var, unsigned int &numCoords) const
void computeMeanAndStandardDeviation (MyNT &mean, MyNT &stddev)
void computeSignificantValues (MyNT pvalueThreshold, MyHistogram &histogram, MyNT &value)
void convertToRanks ()
 Convert the coordinates of every point to their ranks in that point.
void print (string outfile)
 Print the data set to a file.
void print (ostream &ostr)
 Print the data set to an output stream.
virtual void read (MyClusterParams &params, unsigned int whichPointSet=0)
virtual void read (string infile, const MyAffyFileFormat &format=MyAffyFileFormat())
virtual void insert (const MyPoint &point)
virtual unsigned int size () const
virtual unsigned int getNumPoints () const
virtual unsigned int getNumDimensions () const
virtual unsigned int getNumSamples () const
virtual unsigned int getNumGenes () const
virtual bool isPoint (string name) const
virtual vector< unsigned int > getPointIndex (string name) const
virtual const MyPoint & getSample (unsigned int i) const
bool doClassNamesExist () const
string getClassName (int i) const
string getSampleName (int i) const
string getGeneName (int i) const
unsigned int numInClass (string className) const
void setClasses (string className)
void setClasses (const MyClassesType &classes)
unsigned int readClasses (string fileName)
void getClasses (MyClassNamesType &names)
virtual const vector< MyDimInfo > & getDimensionInfo () const
virtual MyDimInfo getDimensionInfo (int i) const
virtual void getDimensionNames (set< string > &names) const
 Return the names of all the dimensions.
virtual void getDimCoords (int dim, vector< MyNT > &coords) const
virtual MyNT getDimensionLow (int i) const
virtual MyNT getDimensionHigh (int i) const
virtual MyNT getDimensionMean (int i) const
virtual MyNT getDimensionSigma (int i) const
virtual void getDimensionValues (unsigned int i, vector< MyNT > &values) const
virtual unsigned int getDimensionCount (unsigned int i, MyNT value) const
virtual const MyPoint & operator[] (int i) const
virtual MyPoint & operator[] (int i)
virtual void getInterval (const vector< int > &discIndices, unsigned int dim, MyNT &low, MyNT &high) const
virtual string suggestClass () const
virtual bool generate (int &index, bool forbidChosen=true) const
virtual bool generate (int num, vector< int > &indices, bool forbidHidden=true, string classToUse="") const
virtual int getNumNotChosen () const
virtual int getNumNotHidden () const
virtual void getPointNames (set< string > &names) const
 Return the names of all the points.
virtual void flip ()
void splitByClass (vector< MyPointSet > &splitPointSets, map< string, unsigned int > &classToIndex)
 If there is more than one class for this point set, split the points by class. If there is just one class, return the point set.
void randomise ()
virtual void choosePoints (const vector< int > &containedPoints)
virtual void hidePoints (const vector< int > &containedPoints)
virtual void hide (const vector< int > &containedPoints, const vector< MyDimensionInterval > &intervals)
virtual MyNT computeEntropy (int dim, MyVerbosityLevel verbosity) const
virtual void computeEntropies (MyVerbosityLevel verbosity)
virtual unsigned int filter (const MyClusterParams &params)
unsigned int filter (MyNT maxDownRegulated, MyNT minUpRegulated)
virtual bool computeCorrelations (string point1, string point2, vector< MyNT > &correlations, string method="Pearsons") const
 Compute the correlations between point1 and point2.
virtual void computeCorrelations (MyNT threshold, string id, bool absoluteValues=true, map< string, map< string, MyNT > > *pairs=NULL)
 Based on the value of threshold, compute statistics on the correlations between all pairs of points and print them to a file whose name is based on id.
virtual void printCorrelations (MyNT threshold, string outputFile, bool absoluteValues=true, map< string, map< string, MyNT > > *pairs=NULL)
 Compute the correlations between all pairs of points and print those greater than a threshold to a file.
MyHistogram printCorrelationsHistogram (string outputFile) const
 Print a histogram of the correlations between all pairs of points to a file.
MyHistogram computeCorrelationsHistogram (map< string, map< string, MyNT > > *pairs=NULL) const
virtual MyHistogram computeRandomisedCorrelationsHistogram (unsigned int numRandomisations, string outputFile)
void standardise ()
void standardise (MyNT m, MyNT s)
void storeRowNameTranslations (const map< string, set< string > > &rowNameAliases)
bool _isFlipped () const

Friends

class MyCluster

Detailed Description

Stores a set of points and information about the point's dimensions. This class is typically useful for storing one gene expression dataset.

Member Function Documentation

void MyPointSet::compareTTest ( string  name,
const MyPointSet &  other,
MyNT &  tstat,
MyNT &  pvalue 
) const

Compare the expression values of point "name" in the invocant and in other by performing a t-test.

Parameters:
[in]name,theidentifier of the point to be compared.
[in]other,theother instance of MyPointSet.
[out]tstat,thevalue of the t-statistic.
[out]pvalue,thep-value of the t-statistic.
Note:
If the point does not exist in either MyPointSet, the resulting tstat and pvalue are 0 and 1, respectively.
MyNT MyPointSet::computeAverageExpressionValue ( const set< string > &  subset)

Compute the average expression value of the points in subset.

The method computes the average expression value of all the points in subset over all the points in the data.

bool MyPointSet::computeCorrelations ( string  point1,
string  point2,
vector< MyNT > &  correlations,
string  method = "Pearsons" 
) const [virtual]

Compute the correlations between point1 and point2.

Compute the correlation between two points.

Parameters:
[in]point1,theID of the first point.
[in]point2,theID of the second point.
[out]correlations,avector of computed correlation values.
[in]method,thetype of correlation to compute. This option can take the following values:

  • Pearsons: compute the Pearson's correlation coefficient.

  • pearsons: same as Pearsons.

  • discrete: compute the discrete correlation coefficient. See the documentation of MyPoint::computeDiscreteCorrelation() for an explanation.

Returns:
true iff the method computed at least one value of correlation.

This method takes the IDs of two points. For every pair of indices corresponding to the points (multiple points may have the same ID), the method computes the correlation between that pair of indices.

Parameters:
[in]point1,nameof the first point.
[in]point2,nameof the second point.
[out]correlations,avector of correlation values.

Given the names point1 and point2 of two points in the invocant, compute the correlations between all pairs of points with these names.

Note:
The reason this method does not return a single value is that in some gene expression data sets more than one gene expression profile may correspond to a single name. It is the job of the caller to combine the multiple correlations returned in a meaningful way.
Warning:
This method runs in time $O(dkl)$, where $k$ is the number of points and $d$ is the number of dimensions in the data. It can take a long time!
Returns:
true, if both points exist in the data and false if either one of them does not.
void MyPointSet::computeCorrelations ( MyNT  threshold,
string  id,
bool  absoluteValues = true,
map< string, map< string, MyNT > > *  pairs = NULL 
) [virtual]

Based on the value of threshold, compute statistics on the correlations between all pairs of points and print them to a file whose name is based on id.

Parameters:
[in]::MyNTthreshold, only print correlations greater than this parameter.
[in]::stringid, a string to construct the name of the file to print the correlations to. If this string is the empty string, the method does not print any correlations.
[in]absoluteValues;if this boolean is true (the default value), the method computes absolute values of the correlations.
[out]pairs,apointer to a 2-dimensional map. If the pointer is not NULL, then the method stores all the pairwise correlations in this map. The method ignores this argument unless the absolute value of thresh is at most 1.
Note:
If you want to obtain a list of pairs of correlations without printing any to an output file, you can pass an empty string for this argument and a non-NULL pointer for the pairs argument.
MyHistogram MyPointSet::computeCorrelationsHistogram ( map< string, map< string, MyNT > > *  pairs = NULL) const

Compute a histogram of the correlations between all pairs of points.

Parameters:
[out]pairs,apointer to a 2-dimensional map. If the pointer is not NULL, then the method stores all the pairwise correlations in this map.
Returns:
An instance of MyHistogram that stores the computed histogram.
void MyPointSet::computeMeanAndStandardDeviation ( MyNT &  mean,
MyNT &  stddev 
)

Compute the mean and standard deviation of the distribution of coordinates in the point set.

bool MyPointSet::computeMeanAndVariance ( string  point,
MyNT &  mu,
MyNT &  var,
unsigned int &  numCoords 
) const

Compute the mean and variance of the coordinates of the average of the vectors corresponding to point.

Parameters:
[in]point,thename of the point.
[out]mu,therequired mean.
[out]variance,therequired variance.
[out]numCoords,thenumber of values used to calculate the mean and standard deviation.
Returns:
true, iff point is an element of the invocant.
Note:
If the mean is 0, variance is 0.
MyHistogram MyPointSet::computeRandomisedCorrelationsHistogram ( unsigned int  numRandomisations,
string  outputFile 
) [virtual]

Randomises the set of points multiple times, computes the correlations between all pairs of points for each random set, and computes a histogram of all the correlations.

Randomises the set of points multiple times, computes the correlations between all pairs of points for each random set, and computes a histogram of all the correlations.

Parameters:
unsignedint numRandomisations, the number of times to randomise the data.
stringoutputFile, the stub of the name of the file to output the histogram to. The default value of this argument is "". The method adds the string "-randomised-num-trials" to outputFile where "num" is the value of numRandomisations. If the file name is provided, the method prints each histogram to the file. A blank line separates two histograms.
Note:
In order to save computation time, if the method notices that if the output file exists, it simply reads the histogram from the output file and returns the histogram. If you want the method to compute the histogram from scratch once more, you must delete the output file.
virtual bool MyPointSet::isPoint ( string  name) const [inline, virtual]

Check if the point name is a point in the invocant.

Parameters:
[in]name,thename of the point.
Returns:
true, if there is a point with the name in the invocant; false, otherwise.
void MyPointSet::printCorrelations ( MyNT  threshold,
string  outputFile,
bool  absoluteValues = true,
map< string, map< string, MyNT > > *  pairs = NULL 
) [virtual]

Compute the correlations between all pairs of points and print those greater than a threshold to a file.

Remarks:
The method ignores points that are hidden, e.g., by MyPointSet::filter().
Warning:
This method runs in time $O(dn^2)$, where $n$ is the number of points and $d$ is the number of dimensions in the data. It can take a long time!
Parameters:
[in]::MyNTthreshold, only print correlations greater than this parameter.
[in]::stringoutputFile, the name of the file to print the correlations to. If this string is the empty string, the method does not print any correlations. The method adds the string "-correlations-threshold-val", where val is the string representation of the parameter "threshold" to the string outputFile.
[in]absoluteValues;if this boolean is true (the default value), the method computes absolute values of the correlations.
[out]pairs,apointer to a 2-dimensional map. If the pointer is not NULL, then the method stores all the pairwise correlations in this map. The method ignores this argument unless the absolute value of thresh is at most 1.
Note:
If you want to obtain a list of pairs of correlations without printing any to an output file, you can pass an empty string for this argument and a non-NULL pointer for the pairs argument.
MyHistogram MyPointSet::printCorrelationsHistogram ( string  outputFile) const

Print a histogram of the correlations between all pairs of points to a file.

Parameters:
stringoutputFile, the name of the file to print the histogram to. The method adds the string "-correlations-histogram" to the name of this file before printing. If the file already exists, the method simply reads the histogram from that file.
Returns:
The computed histogram.
void MyPointSet::randomise ( ) [inline]

Constructs a random permutation of the coordinates of each point.

void MyPointSet::splitByClass ( vector< MyPointSet > &  splitPointSets,
map< string, unsigned int > &  classToIndex 
)

If there is more than one class for this point set, split the points by class. If there is just one class, return the point set.

Parameters:
[out]splitPointSets,avector containing the split set.
[out]classToIndex,amap from class name to indices in the splitPointSets vector.

Friends And Related Function Documentation

friend class MyCluster [friend]
The documentation for this class was generated from the following files:
 All Classes Functions Variables Typedefs Friends
Generated on Tue Jul 31 2012 16:46:50 for Biorithm by   doxygen 1.7.6.1