Reads with high duplication levels and biased gc content

Question

[mjg]

Hi everyone,

I'm working with RNAseq data obtained with a total RNA protocol. It is paired end, and 75 bases long. There is a control and a disease group.

The disease group shows a very consistent (and biased) pattern across samples in terms of GC content and duplication levels (sequences having up to >10,000 K copies). (See images below)

Has anyone seen something similar and found a way on how to explain the issue?

Thanks,
Maria

[Devon Ryan]

If the bias is consistent within the group and doesn't correspond to a batch effect of some sort then perhaps it's real. A high duplication rate is expected in any RNAseq experiment and it's typically very very high in total RNAseq experiments (even if you attempt ribo depletion, you still end up sequencing >10% rRNA half the time).