Recommended Tools For De Novo Motif Discovery In Vertebrate Genome Tsses?

Question

[2184687-1231-83-]

What would be the recommended tools I should try for de novo motif discovery in TSS regions (~1-2KB window around the TSS) of non-model vertebrate genomes?

I've been told I should try CisFinder (http://lgsun.grc.nia.nih.gov/CisFinder), but would like to know of other options for comparison.

Answer 1

[Ryan Dale]

There's been a lot of work recently on MEME to support ChIP-seq-scale data -- see the MEME-ChIP paper and the MEME-ChIP submission form to jump right in.

As part of the standard pipeline, MEME-ChIP also runs DREME, which finds shorter motifs that standard MEME has a hard time with.

I've found MEME-ChIP very easy to use with nice HTML output, and the results I get are consistent with previous work.

It's probably still best to run another algorithm that uses a different strategy -- see this article in Nature Biotechnology for a great intro on the various strategies like enumeration, deterministic optimization, and probabilistic optimization.

Weeder (paper) might be a good choice for this; it was developed specifically for TSSs. However I'm not sure how well it handles big data so you might need to send it random subsamples of your data.

[Alastair Kerr]

It has been over a year since I looked at MEME for ChIP-seq data. I will have another look at it thanks.

Answer 2

[Alastair Kerr]

What type of data do you have?

I do not have any experience with CisFinder. The MEME suite (ref) is commonly used but it can break with 'big data' (sorry I have no quantification on this). We have also had some success with Gimmemotifs for chip-seq data.

[Anima Mundi]

I would suggest the MEME suite too.

Answer 3

[Farhat]

There are a large number of programs that find motifs with different methodologies. This paper compares a number of different programs. The paper is a bit old but most of the algorithms described in it have been updated with the times.