test_genesets_gsea {goat}

GSEA as implemented in the fgsea R package

Description

In most cases, it's more convenient to call the more generic test_genesets function which also deals with multiple-testing correction (per geneset source)

Usage

test_genesets_gsea(
  genesets,
  genelist,
  score_type = NULL,
  parallel_threads = 1L,
  gseaParam = 1,
  nPermSimple = 50000,
  gsea_genelist_col = NULL,
  gsea_scoretype = NULL,
  random_seed = 123
)

Arguments

genesets

data.frame/tibble with geneset and gene columns

genelist

data.frame/tibble with gene and score columns. Should contain columns gene and either pvalue or effectsize, depending on score_type parameter

score_type

how to compute gene scores? options: "pvalue", "effectsize", "custom". Option "pvalue" uses -log10 transformed values from the pvalue column in genelist. Option "effectsize" uses values from the effectsize column in genelist as-is. Option "custom" expects 2 additional parameters; gsea_genelist_col should be a column name in genelist to be used for fGSEA (values used as-is), gsea_scoretype should be the respective value for the fGSEA scoreType parameter ('pos', 'neg' or 'std)

parallel_threads

number of threads to use for parallel processing. Set to 0 to automatically select all available processors/cores, set to 1 to disable (default) or to N to use N processes. Note that multiprocessing sometimes breaks on RStudio + Windows, hence this parameter is set to 1 to disable multiprocessing by default for now

gseaParam

passed to fgsea::fgsea(), from manual: "GSEA parameter value, all gene-level statis are raised to the power of 'gseaParam' before calculation of GSEA enrichment scores.". default = 1. Further comments by fGSEA author at https://github.com/ctlab/fgsea/issues/45

nPermSimple

passed to fgsea::fgsea(), from manual: "Number of permutations in the simple fgsea implementation for preliminary estimation of P-values.". default = 50000 in this R package but 1000 by default in fGSEA v1.22.0; we observed much better accuracy in null simulations when increasing this from default 1k to 10k and further minor improvement towards 50k, hence the latter is our default

gsea_genelist_col

optional, only used for score_type "custom"

gsea_scoretype

optional, only used for score_type "custom"

random_seed

the random seed that is passed to set.seed() in order to ensure fgsea results are reproducible. default: 123

Value

input genesets table with results in the "pvalue", "score_type" and "gsea_nes" columns

See Also

test_genesets