trans.main
Class WilcoxonSignedRankTest

java.lang.Object
  trans.main.WilcoxonSignedRankTest

public class WilcoxonSignedRankTest

extends java.lang.Object

Implementation of a Wilcoxon Signed Rank Test. Many issues to consider: what to do about ties, what to do about zero diffs, what to use as the output (WPlus, WMinus, WCombine, which z-value (Zw or Zs), a p-value, or Affys -10*LogBase10(p-value) transformation, lastly how to handle situations where N<10. The minimal constructor here averages ties, throws out zero diffs, uses the Lowry z-value calculation method and the Affy p-value transformation (transformedPValueFloat) as the final output. Note: this transformedPValueFloat may be 0 indicating no significance (or N<5) (p-value > 0.05), 90 "very" significant (p-value < 1E-11) or something measurable in between. Affys favored 10-5 p-value cutoff corresponds to a LogTrans= 49.4, with bigger numbers "more significant" although in reality, anything bigger than LogTrans 13.1 is, by WSRT, significant, assuming a normal distribution and >10 non zero difference sample pairs. Uncomment the main() to test and tweak.

Constructor Summary

WilcoxonSignedRankTest()

WilcoxonSignedRankTest(float[] treatment, float[] control)
Minimal constructor, fastest way to calculate a transformedPValueFloat

Method Summary

static double	calcSmallSample10Log10PValue(float W, int N) For calculating a -10Log10(pValue) for small sample sizes (N 5-9).
void	calculateSignedRankStatistics() Calculates signed rank statistic S = sum of the ranks (aka WPlus) for WilcoxonSamplePairs with a positive difference and WMinus for those with a negative difference.
void	calculateTestStatisticZs() Calculates test statistic Z from wsignsignrank.pdf Not used in this implementation.
void	calculateTestStatisticZw() Calculates test statistic Z from Lowry .
void	createSamplePairs(float[] treatment, float[] control, boolean noZeros) Creates WilcoxonSamplePair objects from two float[]s.
double[]	getAverageIntensityValues()
int	getN()
int	getNumberWMinusRanks()
int	getNumberWPlusRanks()
double	getTransformedPValue()
java.lang.String	getTransformedPValueString()
float	getW()
float	getWMinus()
float	getWPlus()
WilcoxonSamplePair[]	getWsps()
float	getZs()
float	getZw()
static void	main(java.lang.String[] args)
void	rankSamplePairs() Ranks a sorted array of WilcoxonSamplePairs based on absoluteDifference.
void	setWsps(WilcoxonSamplePair[] pairs)
static float[]	signedDifference(float[] scores, float median) Returns signed difference between the scores and a median.
double	test()
java.lang.String	toString()

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

WilcoxonSignedRankTest

public WilcoxonSignedRankTest(float[] treatment,
                              float[] control)

Minimal constructor, fastest way to calculate a transformedPValueFloat

WilcoxonSignedRankTest

public WilcoxonSignedRankTest()

Method Detail

test

public double test()

calculateTestStatisticZw

public void calculateTestStatisticZw()

Calculates test statistic Z from Lowry .

calculateSignedRankStatistics

public void calculateSignedRankStatistics()

Calculates signed rank statistic S = sum of the ranks (aka WPlus) for WilcoxonSamplePairs with a positive difference and WMinus for those with a negative difference.

rankSamplePairs

public void rankSamplePairs()

Ranks a sorted array of WilcoxonSamplePairs based on absoluteDifference. If no ties are found then this is simply their array index number+1. (ie 1,2,3,4...) If ties are encountered, ties are assigned the average of their index positions+1. (ie if index+1's: 2,3,4 have the same absolute difference, all are assigned a rank of 3).

createSamplePairs

public void createSamplePairs(float[] treatment,
                              float[] control,
                              boolean noZeros)

Creates WilcoxonSamplePair objects from two float[]s. Indicate if you want to throw out pairs with a difference of zero. (true for no zeros, false for keep zeros

signedDifference

public static float[] signedDifference(float[] scores,
                                       float median)

Returns signed difference between the scores and a median.

calcSmallSample10Log10PValue

public static double calcSmallSample10Log10PValue(float W,
                                                  int N)

For calculating a -10Log10(pValue) for small sample sizes (N 5-9).

toString

public java.lang.String toString()

getWsps

public WilcoxonSamplePair[] getWsps()

setWsps

public void setWsps(WilcoxonSamplePair[] pairs)

getZw

public float getZw()

getNumberWMinusRanks

public int getNumberWMinusRanks()

getNumberWPlusRanks

public int getNumberWPlusRanks()

getW

public float getW()

getWMinus

public float getWMinus()

getWPlus

public float getWPlus()

getTransformedPValue

public double getTransformedPValue()

getTransformedPValueString

public java.lang.String getTransformedPValueString()

calculateTestStatisticZs

public void calculateTestStatisticZs()

Calculates test statistic Z from wsignsignrank.pdf Not used in this implementation.

main

public static void main(java.lang.String[] args)

getN

public int getN()

getZs

public float getZs()

getAverageIntensityValues

public double[] getAverageIntensityValues()