Looking for some guidance regarding motif ontology

Question

[Ronin]

Hi all,

I'm working on an ATAC-seq project, and I recently ran some de novo motif finding (HOMER, findmotifsgenome.pl) and am happy with what results I have, but I would like to go farther. Being able to understand the biological significance of these motifs seems like the logical next step.

I was wanting to see if anybody here knew of some pipelines that involve using HOMERs findMotifs or findMotifsGenome to use as inputs for some type of ontology work. Alternatively, if there are some other motif finding and ontology pipelines that are recommended, I am more than willing to try them. I am, however, working with a non-model species, and so perhaps my resources are more limited, but I'm hardly an expert here.

Cheers,

Answer 1

[jared.andrews07]

The MEME suite has a huge variety of tools available for de novo motif identification, comparison to know motifs, motif enrichment, etc. They also have a very comprehensive array of databases available.

It is also regularly updated. I'd highly recommend at least giving it a shot, as most functionality is available via their web server.

[Ronin]

Thanks for the recommendation. I've been trying to run that for about a day and a half on my merged peak file. I'm unsure if it is supposed to take this long, or if I've made some mistake. So far what I've done is:

1) use macs2 'callpeak' to generate my peak file 2) convert the bed file to fasta using bedtools 3) run meme on the peak file (meme-chip -oc motif_results -meme-nmotifs 200 ./macs2.liver_ATAC_peaks.narrowPeak.fa)

It's been about 36 hours since I ran the script, and it isn't completed yet. I'm fairly convinced I've made some mistake here, but I can't find any tutorials that walk you through the steps and preparation of files. Do you have any suggestions?

[jared.andrews07]

I've actually never ran that specific tool, but no, I don't remember things taking that long. Requesting 200 motifs is a ton though, and I expect that's what is taking so long. Usually you'd request like <10.

I would not run MEME-ChIP given you have ATAC peaks, which aren't very large anyway. Just run MEME, optionally followed by TOMTOM if there's an applicable database for your organism. MEME is parallelizable.