pSI (version 1.1)

fisher.iteration: Fisher's Exact Test Across All Cell Types & pSI Thresholds

Description

fisher.iteration will test a candidate gene list for overrepresenation in the various cell type/pSI threshold combinations produced by the specificty.index function. NOTE:Supplementary data (human & mouse expression sets, calculated pSI datasets, etc.) can be found in pSI.data package located at the following URL: http://genetics.wustl.edu/jdlab/psi_package/

Usage

fisher.iteration(pSIs, candidate.genes, background = "data.set", p.adjust = TRUE)

Arguments

pSIs

data frame output from specificity.index function with the number of columns equal to the number of samples and genes as rows.

candidate.genes

candidate gene list tested for overrepresentation in cell types/samples. Comprised of official gene symbols.

background

character string used to indicate what background gene list should be used in Fisher's exact test for overrepresentation. The default value is "data.set" which indicates that the gene list of the input pSI data set will be used to represent the background gene list. This would be used in the case when the input pSI data set is comprised of genes derived from the same species as the genes found in the candidate gene list. background can take on two other values, the first of which is "human.mouse". "human.mouse" indicates that the background gene list will be comprised of intersection of two lists: 1) all genes in the input pSI dataset (all are human genes), 2) all genes with clear human-mouse homologs. This option would be used in the case when the input data set is comprised of human genes (i.e. genes from a human microarray) and the candidate gene list being tested is comprised of mouse genes. The last value background can take on is "mouse.human". "mouse.human" indicates that the background gene list will be comprised of intersection of two lists: 1) all genes in the input pSI dataset (all are mouse genes), 2) all genes with clear mouse-human homologs. This option would be used in the case when the input data set is comprised of mouse genes (i.e. genes from a mouse microarray) and the candidate gene list being tested is comprised of human genes.

p.adjust

logical. default output is bonferroni corrected p-value but if p.adjust is FALSE, nominal p-values will be output.

Details

This function is used to answer the question of what is the probability that a certain number of genes specific to a certain cell type/sample occured by chance (as usual with low probabilities corresponding to high statistical significance). This is accomplished with a binary variable for each gene in the population with two mutual exclusive values: 1) The gene is specific to the cell type/sample in question or 2) The gene is not specific to the cell type/sample in question

Examples

##load sample pSI output
data(sample.data)
##load sample candidate gene lists
data(candidate.genes)
##run Fisher's exact test for overrperesentation on pSI.out for the AutDB
##candidate gene list across all cell types/sample types & pSI thresholds
fisher.out.AutDB <- fisher.iteration(pSIs=sample.data$pSI.output,
                                         candidate.genes=candidate.genes$AutDB)