Hi, I’m comparing enriched motifs between two related region sets (low vs high chipseq peaks). For HOMER, is it preferable to use the automatic background for each set independently, or to explicitly use one set as the background for the other? How robust is HOMER’s automatic background in general?
Thanks so much for the help!
1 answer
They answer different questions. Against the automatic background you'll get the same master motif coming up in both sets and it'll look like nothing distinguishes them - using one set as background for the other is what actually asks "what's different about the strong peaks". I'd run it both directions.
One thing to fix before you do though: use the same region size for both (-size 200 around the summit rather than -size given). Peak strength tracks peak width and GC content, so otherwise you're partly reading that rather than motif biology. HOMER's GC normalisation is reasonable but it can't undo a width difference.
Log in to answer this question.