I have a query about ChIP-Seq data. I have been working for sometime with ChIPSeq single end data and till date I have worked for samples with 30X, 40X and 60X coverage data. I have worked for few transcription factors and histone marks and I have always found the samples after alignment having more than 20 million reads (uniquely mapped). Today I received some new samples from patients for array of Chromatin marks where the H3K4me1, H3K4me3 and H3K27ac are sequenced with a coverage of 30X and H3K27me3 and inputs are done with a coverage of 60X. Looking at the FasQC report I found that some samples have less than 20 million reads. So after mapping it will farther drop. If I look at ENCODE and other ChIPSeq guidelines , most mention about having 20 million mapped reads to have better peak calling and enrichment. Infact we are interested in enchancers (active, primed and poised). So should I ask the facility to re-sequence the samples for which the reads are below 25 million (since on mapping they will go lesser than 20 million mapped reads, this am saying as a part of preliminary analysis for this same project I got a mapping median of 78% on previous set of samples) ? Another thing our facility said
" we can reduce the coverage and pool together the reads of the two runs"
I actually did not get what does this mean. These are samples from patients run for 4 different marks. Should not we have uniform coverage for all the samples to start with although each sample has different number of reads and also mapping will differ. But lowering coverage , is it advisable? If we started with 30X and 60X , it should be same for all samples right? Unless I run any assessment of QC on the aligned reads and then make peak calling for the low quality samples I will not be able to make any conclusion. The re-sequencing comes with a cost. So I would like to ask from the community experts , first what did the above comment from our facility mean. Also should you all think I go ahead with analysis from the samples I already received? Any suggestions and discussion are appreciated