Variable gap motif

Question

[Satyajeet Khare]

Hi Biostars,

I am performing ChIP-Seq analysis of a transcription factor that binds as a dimer to palindromic sequences with variable gap in between. I need to align these motifs with different gap lengths back onto the genome in order to identify direct targets of the TF. In order to do this, I tried MEME-ChIP and Glam2 with default settings so far. I do see a palindromic motif in output which is quite similar to known motif. I can also see that the motif is a variable gapped by manual inspection.

But If I align this motif back to the genome using FIMO, I see many sequences at ChIP-peaks that match the motif but do not get picked up by FIMO.

Since there is a clear ChIP peak on these sequences and there is a clear motif sequence at the summit, I am baffled why such sites are not getting picked by genome alignment by FIMO. The sites picked by FIMO also show clear motifs indicating FIMO has worked too.

Am I missing something? Is there a better tool or different way to work with gapped motifs? Thanks for your help!

P.S. I have also tried SICER for gapped motifs.

Answer 1

[Santosh Anand]

FIMO uses p-value cut-off of 1e-4, try lowering that (Advanced options on web). On the other hand, FIMO might not be good for a gapped motif. The MEME website suggests GLAM2SCAN for that.

[Satyajeet Khare]

Thanks Santosh,

Actually FIMO was not reporting all scores. So I had to play with --max-stored-scores and increase it to 500000. To my surprise, even at higher stringency (p-value 5e-5), I started to see alignment at expected motifs on the genome.