mpmeers
commented
5 years ago Hi Lauren,
I can't speak to MACS2 bdgdiff, but collaborators of ours have used DiffBind in combination with SEACR with some success--in that case you would either have to establish a common peak set between treatment and control, perhaps by taking the intersection of peak lists using bedtools intersect or bedops -i, or use DiffBind dba function on a sample sheet. My understanding is that DiffBind will map reads to the common peak set and use a DESeq-like geometric normalization strategy to derive differential binding calculations, meaning the total signal column from SEACR wouldn't be necessary. I personally have not used DiffBind though, so I recommend you read the documentation carefully (http://bioconductor.org/packages/release/bioc/vignettes/DiffBind/inst/doc/DiffBind.pdf) to make sure what I'm saying is accurate.
As an alternative you might be able to use the total signal column itself as a quick qualitative measure of differential binding, provided that A) the peaks that you are comparing in treatment vs. control are almost entirely overlapping, and B) your bedgraph signal is normalized in a way that reflects biological differences between sample, e.g. normalized to some sort of spike-in that's common between treatment and control.
Let me know if any of that works for you, and good luck,
Mike
Hi!
I've run SEACR successfully on samples that contain a treatment and control. I'm looking for ways to now compare the peaks between these conditions with the SEACR output. Do you have a recommendation for a program? I've seen recommendations for DiffBind or MACS2 bdgdiff. Would I use the total signal column of SEACR output as the score for these programs? Thanks for any advice!
Sincerely, Lauren