Centrimo Plot interpretation

[peter.mc...@gmail.com]

Hi MEME Team

I've been using MEME-ChIP to analyse a 150bp region around some H3K27ac peaks we have identified.

 

In the Centrimo output I see TFs that are centrally enriched but then others who 'dip' around the centre and the become enriched in the flanking regions (example attached). I ran Centrimo in local mode.

What is causing some TFs to exhibit this 'dip'? Can I say that they are enriched around my peak, but just not centrally? Is this an indication that some TFs (e.g. STAT2) usually bind in the centre but have somehow been displaced?

Kind Regards,

Peter

[CharlesEGrant]

Notice that the motifs for MECP2 and SP4 are GC rich, while the motifs for MNX1 and STAT2 are AT rich. That means their enrichment scores are always going to be anti-correlated. It may be that Centrimo is just reporting the GC-bias in your sequences. I ran your results by Tim Bailey, the author of Centrimo and MEME-ChIP. He was a bit concerned that your peak were so sharp, and so perfectly centered. Did you center your sequences on some landmark, like a known motif location? This would conflict with Centrimo’s null model assumption that the sequences are independent, and motifs uniformly distributed within them.