TFBS ChIP-Seq: No Cross-correlation Peaks other than Read Length: Is it normal?

Question

[John Ma]

I'm currently analyzing someone else's ChIP-seq data, and noticed after routine mapping and filtering, there's only one real peak from the cross-correlation plot generated by phantompeakqualtools at read length. While PPQ tools also called other peaks, they're obviously artifacts. NSC is within normal range (5-10), RSC is low (0.1-0.3), but this can be excused by the low expected numbers of TFBS.

In this case, is it appropriate to go on with peak calling?

Answer 1

[Devon Ryan]

NSC/RSC and strand cross-correlation are mostly meaningless, don't base anything off of them (or even bother calculating them). A more useful thing to do is look at a "fingerprint" plot (e.g., from plotFingerprint in deepTools), though really you're going to do peak calling anyway regardless of what the results are, since that's quick and repeating experiments takes a while.