Package org.snpeff.probablility

Class RankSumNoReplacementPdf

java.lang.Object

org.snpeff.probablility.RankSumNoReplacementPdf

public class RankSumNoReplacementPdf
extends java.lang.Object

Calculate rank sum probability distribution function (pdf) and cumulative distribution function (cdf). Note: This class assumes that ranks cannot be repeated (selecting without replacement)

Author:

Pablo Cingolani

Field Summary

Fields

Modifier and Type	Field	Description
static int	CACHE_MAX_N	Cache size roughly (N * NT)^2
static java.lang.String	DEFAULT_CACHE_FILE
static int	warnCDF

Method Summary

Modifier and Type	Method	Description
void	cacheSetPdf(int n, int nt, long r, double pdf)
boolean	canBeCached(int n, int nt)	Is the number in the cache
org.apfloat.Apfloat	cdf(int n, int nt, long r)	Probability of getting a rank sum less or equal to 'r' when adding the ranks of 'nt' selected items.
org.apfloat.Apfloat	cdfExact(int n, int nt, long r)	Cumulative density function (cdf)
org.apfloat.Apfloat	cdfNormal(int n, int nt, long r)	Normal approximation to rankSum CDF
org.apfloat.Apfloat	cdfTriangle(int n, int nt, long r)	Uniform 'approximation' to rank sum statistic
org.apfloat.Apfloat	cdfUniform(int n, int nt, long r)	Uniform 'approximation' to rank sum statistic
void	createCacheFile()	Create a cache file
static RankSumNoReplacementPdf	get()
java.lang.String	getCacheFile()
long	maxRankSum(int n, int nt)	Maximum possible rank sum
double	mean(int n, int nt)	Mean value for a given N and N_T
long	minRankSum(int n, int nt)	Minimum possible rank sum
org.apfloat.Apfloat	pdf(int n, int nt, long r)	Probability of getting a rank sum equal to 'r' when adding the ranks of 'nt' selected items.
org.apfloat.Apfloat	pdfExact(int n, int nt, long r)	Probability of getting a rank sum equal to 'r' when adding the ranks of 'nt' selected items.
org.apfloat.Apfloat	pdfExact(int n, int nt, long r, long rmin, int out)	Probability density function (pdf): Exact calculation
org.apfloat.Apfloat	pdfNormal(int n, int nt, long r)	Normal approximation to rank sum statistic
org.apfloat.Apfloat	pdfTriangle(int n, int nt, long r)	Uniform 'approximation' to rank sum statistic
org.apfloat.Apfloat	pdfUniform(int n, int nt, long r)	Uniform 'approximation' to rank sum statistic
void	readCacheFile()	Read cache file
void	setCacheFile(java.lang.String cacheFile)
double	sigma(int n, int nt)	Wrapper to Sqrt(variance)
java.lang.String	toStringCache()	Some cache statistics
double	variance(int n, int nt)	Variance for a given N and N_T
double	varianceSlow(int n, int nt)	Variance for a given N and N_T (slow method, only used for debugging)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

CACHE_MAX_N

public static int CACHE_MAX_N

Cache size roughly (N * NT)^2

DEFAULT_CACHE_FILE

public static java.lang.String DEFAULT_CACHE_FILE

warnCDF

public static int warnCDF

Method Detail

get

public static RankSumNoReplacementPdf get()

cacheSetPdf

public void cacheSetPdf(int n,
                        int nt,
                        long r,
                        double pdf)

canBeCached

public boolean canBeCached(int n,
                           int nt)

Is the number in the cache

Returns:

true if it is

cdf

public org.apfloat.Apfloat cdf(int n,
                               int nt,
                               long r)

Probability of getting a rank sum less or equal to 'r' when adding the ranks of 'nt' selected items. Items are ranked '1..n' (from 1 to 'n') Note: Approximated when 'n > 30'

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

cdfExact

public org.apfloat.Apfloat cdfExact(int n,
                                    int nt,
                                    long r)

Cumulative density function (cdf)

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is less or equal to 'r'

cdfNormal

public org.apfloat.Apfloat cdfNormal(int n,
                                     int nt,
                                     long r)

Normal approximation to rankSum CDF

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is less or equal to 'r'

cdfTriangle

public org.apfloat.Apfloat cdfTriangle(int n,
                                       int nt,
                                       long r)

Uniform 'approximation' to rank sum statistic

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

dr - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is equal to 'r'

cdfUniform

public org.apfloat.Apfloat cdfUniform(int n,
                                      int nt,
                                      long r)

Uniform 'approximation' to rank sum statistic

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is equal to 'r'

createCacheFile

public void createCacheFile()

Create a cache file

Parameters:

fileName -

getCacheFile

public java.lang.String getCacheFile()

maxRankSum

public long maxRankSum(int n,
                       int nt)

Maximum possible rank sum

Parameters:

n -

nt -

Returns:

mean

public double mean(int n,
                   int nt)

Mean value for a given N and N_T

Parameters:

n -

nt -

Returns:

minRankSum

public long minRankSum(int n,
                       int nt)

Minimum possible rank sum

Parameters:

n -

nt -

Returns:

pdf

public org.apfloat.Apfloat pdf(int n,
                               int nt,
                               long r)

Probability of getting a rank sum equal to 'r' when adding the ranks of 'nt' selected items. Items are ranked '1..n' (from 1 to 'n') Note: Approximated when 'n > 30'

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

pdfExact

public org.apfloat.Apfloat pdfExact(int n,
                                    int nt,
                                    long r)

Probability of getting a rank sum equal to 'r' when adding the ranks of 'nt' selected items. Items are ranked '1..n' (from 1 to 'n') Note: Exact calculation Wrapper to 'real' pdf function

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

pdfExact

public org.apfloat.Apfloat pdfExact(int n,
                                    int nt,
                                    long r,
                                    long rmin,
                                    int out)

Probability density function (pdf): Exact calculation

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is equal to 'r'

pdfNormal

public org.apfloat.Apfloat pdfNormal(int n,
                                     int nt,
                                     long r)

Normal approximation to rank sum statistic

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is equal to 'r'

pdfTriangle

public org.apfloat.Apfloat pdfTriangle(int n,
                                       int nt,
                                       long r)

Uniform 'approximation' to rank sum statistic

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

dr - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is equal to 'r'

pdfUniform

public org.apfloat.Apfloat pdfUniform(int n,
                                      int nt,
                                      long r)

Uniform 'approximation' to rank sum statistic

Parameters:

n - : Maximum rank number

nt - : Number of elements in the sum

r - : rank sum value

Returns:

The probability that selecting 'nt' elements out of 'n' ranked elements, the rank sum is equal to 'r'

readCacheFile

public void readCacheFile()

Read cache file

setCacheFile

public void setCacheFile(java.lang.String cacheFile)

sigma

public double sigma(int n,
                    int nt)

Wrapper to Sqrt(variance)

Parameters:

n -

nt -

Returns:

toStringCache

public java.lang.String toStringCache()

Some cache statistics

Returns:

variance

public double variance(int n,
                       int nt)

Variance for a given N and N_T

Parameters:

n -

nt -

Returns:

varianceSlow

public double varianceSlow(int n,
                           int nt)

Variance for a given N and N_T (slow method, only used for debugging)

Parameters:

n -

nt -

Returns: