goseq_wrapper.R

#Wrapper around goseq 
#'@param assayedGenes total gene IDs that were measured
#'@param deGenes DE gene IDs
#'@param source_id Can be `ensGene` or `geneSymbol`
#'@param  hyperGeo Dfault TRUE. Set to FALSE for rna-seq data
goseq_wrapper = function(assayedGenes, deGenes, source_id = "ensGene", hyperGeo = TRUE){
  gene_vector = as.integer(assayedGenes %in% deGenes)
  names(gene_vector)= assayedGenes
  
  pwf = suppressWarnings(suppressMessages(goseq::nullp(DEgenes = gene_vector, genome = "hg19", id = source_id, plot.fit = FALSE)))
  if(hyperGeo){
    GO = suppressWarnings(suppressMessages(goseq::goseq(pwf = pwf, genome = "hg19", id = source_id, test.cats = c("GO:BP", "GO:MF"), method = "Hypergeometric")))
  }else{
    GO = suppressWarnings(suppressMessages(goseq::goseq(pwf = pwf, genome = "hg19", id = source_id, test.cats = c("GO:BP", "GO:MF"))))
  }
    
  data.table::setDT(GO)
  GO[,fdr := p.adjust(GO$over_represented_pvalue)]
  GO
}