Recommendations for a good motif-finder, for short K-mers?

Question

[blur]

Hi, I am trying to find a motif in a few thousand 100bp long sequences, and I think the motif I am trying to find is very short (a few bp, perhaps even 3-4bp long) I tried MEME and HOMER, but didn't find any significant motifs. Any recommendations?

[Lars Juhl Jensen]

Significant in which sense? Compared to another set of sequences? Compared to some background model? Do you expect the motif to occur multiple times in each sequence or just once? Would you expect the motif to be evolutionarily conserved?

[blur]

whatever way I tried, I seem to be at a loss - HOMER I tired with background datasets, with different datasets... nothing. I get very few possible motif option with p-values that are just not reliable... I thought perhaps a short motif was too short for these algorithms? As for the motif itself, I have a suspicion it would be something like GC-rich, 4bp sequence, as similar analyses to what I do seem to end up like that, but I don't know if I am correct... Any insight would be greatly appreciated

Answer 1

[le2336]

Are you trying to perform a de novo or targeted motif search? For the former, DREME is better suited for short (<8 positions) motif discovery. If you already know what k-mers you want to search for, you might want to perform a targeted motif search with FIMO to improve your sensitivity.