Hi all,

I was trying to use repenrich2 to compare the amount of reads from H3K9me3 ChIP that fits repeat annotation in two different cell line(mutant vs WT). Repenrich2 uses both genome coordinate annotated in repeat masker, as well as the canonical sequence of the repeats the identify reads that belongs to different repeat classes. Sample output format of repenrich2 All samples(both IP and input) has the same number of total read count.

During this process, I found that the input of mutant cells has a lot more reads with the annotation of repeat type A. I have read the following threads Modeling input data in ChIP-seq experiments with EdgeR, DESeq(2) or Limma and Question: DESeq2 for ChIP-seq differential peaks and it seems like the general consensus is that normalizing against input is not recommended and those regions should be masked out. However, I think it might be interesting to take input into account since:

1. It is very possible that this is a biologically significant finding, as repeat expansion had been previously observed with this mutation.

2. As suggested by the first thread I linked, it suggested that the change in input might be associated with change in chromatin state. However, if novel insertion site of repeats are found, then should we take input into account?

Thank you so much in advance.

I read, but don’t have any particular opinion or input on this question.