GRS: Function to calculate the Generalized Random Set statistic to...
In uc-bd2k/CLEAN: Clustering Enrichment Analysis

View source: R/GRS.R

GRS	R Documentation

Function to calculate the Generalized Random Set statistic to perform Functional Annotation by Primary Genomics Data

Description

This function is intended to compare a new genomics dataset to existing primary genomics data by using the Generalized Random Set approach to compute the concordance of gene lists based on corresponding p-values

Usage

GRS(query.p, reference.p, query.gL = NULL, reference.gL = NULL, na.rm = TRUE, estimateNullDistr = TRUE, nullDistrQuantiles = c(0.9, 0.95, 0.99), nullDistrN = 100, tolerateWarnings=TRUE, pAdjust.method.query=NULL, pAdjust.method.reference=NULL, lambda=0.005)

Arguments

`query.p`	Query p-values. A vector, a matrix with two columns where the first column contains the geneIDs and the second column contains the p-values, or a matrix with three columns where the first column contains the geneIDs, the second contains the scores and the third contains probability. If the query.p is a vector and query.gL is NULL, query.p and reference.p are assumed to have the same number of genes and genes are in the same order.
`reference.p`	Reference p-values. A vector, a matrix with two columns where the first column contains the geneIDs and the second column contains the p-values, or a matrix with three columns where the first column contains the geneIDs, the second contains the scores and the third contains probability. If the reference.p is a vector and reference.gL is NULL, query.p and reference.p are assumed to have the same number of genes and genes are in the same order.
`query.gL`	If query.p is a vector of p-values, this parameters is expected to contain a vector of corresponding gene IDs of same length. Otherwise it is ignored.
`reference.gL`	If reference.p is a vector of p-values, this parameters is expected to contain a vector of corresponding gene IDs of same length. Otherwise it is ignored.
`na.rm`	If TRUE, any NA's in query.p and/or reference.p are removed before computing the GRS score.
`estimateNullDistr`	If TRUE, an empirical distribution of the gene-specific GRS score is estimated.
`nullDistrQuantiles`	If 'estimateNullDistr' is TRUE, specifies the desired quantiles of the null-distribution.
`nullDistrN`	If 'estimateNullDistr' is TRUE, specifies the number of iterations for the estimation function.
`tolerateWarnings`	If FALSE, function returns (with the warning message as character string). This parameter is intended when using GRS() by the Genomics Portals.
`pAdjust.method.query`	If NULL, the probability of differential expression is set to posterior probability of diff expression based on query.p, else 1-adjusted pvalue is used from the p.adjust function.
`pAdjust.method.reference`	If NULL, the probability of differential expression is set to posterior probability of diff expression based on reference.p, else 1-adjusted pvalue is used from the p.adjust function.
`lambda`	To be passed to qvalue() function which estimates the proportion of non-differentially expressed genes.

Details

Value

This function returns a list:

`p.value`	GRS p-value. Under the null-hypothesis, the two gene lists are not concordant.
`z.score`	GRS z-score
`geneTable`	Table where the first column contains the common gene IDs and the second column contains the geney-specific GRS score.
`EvalueNullDistrQ`	If 'estimateNullDistr' is TRUE, this list item gives the quantiles of the null-distribution specified by 'nullDistrQuantiles' parameter.

Author(s)

Johannes Freudenberg

References

Publication forthcoming.

Examples

 data(bCa)
 res <- GRS(bCa$Schmidt.pValues, bCa$Miller.pValues)

 data(gimmOut)
 p <- apply(gimmOut$clustData[,-(1:2)], 1, function(x) t.test(x[1:4], x[5:8])$p.val)
 p <- cbind(geneIDs=gimmOut$clustData[,1], p=p)
 p2 <- convertGeneTable(p, from="r", to="h")
 GRS(bCa$Schmidt.pValues, p2)

uc-bd2k/CLEAN documentation built on Sept. 22, 2022, 4:12 a.m.