Hello everyone,

We have recently performed ChIP-seq experiments against TFs and Chromatin Marks and the sequencing files appeared to have very strong bias on A/T (around 40% each) against G/C (around 10% each). The overall sequencing quality is very poor and we had an extreme overrepresentation of poly-AT 50mers (and their variations, eg. ATATATA... , GATATATATAT.., TATATATAT..., GTATA.., etc.) which covered around 20-30% of the reads. This "wierd" overrepresentation was abserved more or less in all of our samples (4 lanes), to a lesser or greater extend. However, two lanes included samples that were processed with the same library prep method as the above gave positive results, although one of them had and the other one lacked this overrepresentation. We asked the sequencing facility and the machine doesn't seem to have an issue.

Given the above, does anybody have an idea where do these poly-AT mers come from? We are trying to optimize our method but we are not sure, is it an actual ChIP issue (non-specific binding to low complexity regions)? Is it a library prep/contamintation issue?

Our samples were run on Illumina Hi-Seq 2500, 50bp single end.

Many thanks, Nikos T.