Questions about GimmeMotifs output

[paumarc]

I have been using gimmemotifs to analize data comming from chipseq experiment, it worked well, but now is hard for me to understand the outputs and i miss some information at the documentation. Let me explain myself:

at the results i have four html files : gimme.clustereds.html, gimme.denovo.html, gimme.motifs.html, gimme.motifs.redundant.html

that i think, but not for sure, that contains:

gimme.denovo.html : motives found that are not present at any datbase
gimme.motifs.redundant.html : All motives found at the sample, including denovo and known motif
gimme.clustereds.html here i think you try to "cluster" the motives
gimme.motifs.html : Final results with the most relevant information found

is that rigth?

the point here is that i don't find how to how to follow the the process, should I find the names from gimme.clustereds.html in gimme.motifs.redundant.html ?

how do you go from them redundant list to the motif list? and what should i take as a final result

thanks

[simonvh]

Hi @paumarc, indeed, the documentation is not complete.
All in all, you are correct in your description of the files. The most important file is gimme.motifs.html. This provides an overview of the most important motifs, while trying to take care of motif redundancy. This includes (by default) both known as well as de novo motifs. However, this process of removing redundancy is not always perfect, so sometimes it's good to look at the gimme.motifs.redundant.html file to see if there's anything notable that's missing from the final output. In addition, if you want to compare between different runs, you have to use the gimme.motifs.redundant.html, as the grouping in gimme.motifs.html is run- and result-dependent.

The file gimme.denovo.html gives a more detailed overview of the quality of the de novo motifs. These motifs are clusered, as gimme runs a whole load of motif prediction tools with various settings. These are the ones that are present in gimme.clustereds.html. I pretty much never look at this output, as it contains many, many similar motifs.

Hope this helps!