Chip-Seq Of Conventional Tfs: What Are The Peaks W/O The Motif?

Question

[predeus]

Hello all,

I've read several papers estimating the quality of ChIP-seq and most estimate percentage of peaks that include the consensus sequence at about 50%. Having lower FDR does not improve this very dramatically (to 70% maybe).

So what are the high confidence peaks that do not contain the target motif? What causes them? Is there a systematic study of any sort?

I'd be grateful for any pointers.

cheers

Answer 1

[Mikael Huss]

Some possible explanations could be:

1. Secondary binding/protein complexes: Another protein is binding the DNA and your ChIP:ed protein is binding to that protein.
2. Generally open chromatin: an active regulatory region (perhaps "unlocked" by a "pioneering TF") is easier to bind to in the absence of a motif.
3. Degenerate motif: there is actually a binding motif but it is relatively dissimilar from the consensus
4. Artefactual regions, see e g the "black list" here: https://sites.google.com/site/anshulkundaje/projects/blacklists

Since you have an FDR it seems you have a control library so the following may not apply to your case but is interesting in general: even in negative control libraries, regions like transcription start sites give robust peaks: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0005241

Answer 2

[predeus]

Great. Thank you for the answer and the reference.