gsea_null-methods: GSEA null model

In fossbert/binilib: Carlo's little helper

Description

This function generates a GSEA null distribution of enrichment scores based on gene shuffling, i.e. the positions of genes in the signature are permuted rather than phenotype labels. It will add a normalized enrichment score (NES) to the gsea object as described by Subramanian et al. and statistical inference is carried out using one-tailed tests on the appropriate (positive/negative) side of the null distribution. For two-tailed GSEA, the null model is calculated accordingly and separate NES and p-values are provided for each gene set.

Usage

gsea_null(gsea_obj, ...)

## S4 method for signature 'gsea1'
gsea_null(
  gsea_obj,
  w = 1,
  perm = 1000,
  seed = 42,
  analytical = FALSE,
  verbose = TRUE
)

## S4 method for signature 'gsea2'
gsea_null(gsea_obj, w = 1, seed = 42, perm = 1000, verbose = TRUE)

Arguments

gsea_obj	a gsea object as returned by gsea1T or gsea2T/gsea_regulon
...	Additional parameters added to keep compatibility
w	numeric indicating the weights used for GSEA (defaults to 1)
perm	integer indicating the number of permutations to carry out (defaults to 1000)
seed	integer for random number generation during the permutation process (defaults to 42)
analytical	logical whether to derive p-values analytically, i.e. from two-tailed test using a normal distribution (defaults to FALSE, i.e. inference according to Subramanian et al.)

Value

gsea_obj with information on nullmodel added in extra slots

fossbert/binilib documentation built on April 23, 2021, 10:31 p.m.