To perform Over Representation Analysis (ORA) firstly the <*hypergeom_gene_set_databases*> database was filtered to include only genes that are also in both the normalised expression file and the background file (the expression set). Next the database was filtered to include only gene sets with between <*hypergeom_gene_set_min_set_size*> and <*hypergeom_gene_set_max_set_size*> valid genes. For each set of candidate genes (i.e. each signature) the number of overlapping genes with each gene-set in the database was determined. Next a hypergeometric test was performed for each candidate set and gene set combination using the number of genes in the expression set as the background size. To correct for multisampling a Benjamini-Hochberg correction was applied. Gene sets with an adjusted p-value of <*hypergeom_gene_set_padj*> and an absolute log2fold enrichment above <*hypergeom_gene_set_log2fold*> were considered significant.
