From a list of 3000 significantly differentially expressed genes (from a universe of 20000 genes using RNAseq gene expression data), I used DAVID Functional Annotation to extract pathway and gene ontology enrichment. DAVID came up with GO_BP_FAT results using 73% of the genes, but the KEGG_PATHWAY output had only 5 pathways (p-value < 0.05) enriched using only 32% of the gene list.
What could cause this discrepancy in coverage between GO enrichment and Pathway enrichment? Any ideas on how to get better coverage from DAVID for kegg pathways? Or, are there superior platforms for deriving kegg enrichment? I was thinking about incorporating protein-protein interactions to build my gene list, but at that point I'd get too close to the total gene universe.