I'm having some troubles both estimating the quality and understanding the results of my ChIp-Seq samples as well as to decide how to continue. I hope I can find some help/advice here.

I have three conditions, each with duplicates for IP and Input (in total 12 samples). When running fastqc on the samples I already noticed a high duplication rates for all samples, which is why I have tried to get rid of them, when mapping the samples using the STAR aligner with the parameters --outFilterMultimapNmax 1 --alignIntronMax 1 --outFilterMismatchNmax 1 --alignEndsType EndToEnd.

Converting the resulted bam files into bigwig for visualization (no normalization was done yet) is shown in the image below (using igv tools).

I want now to use deepTools to compare the samples and have tested these with the plotFingerprints and the bamCorrelation plots but I can see very strange behavior of some of the IPed samples. There is almost no difference between them and the input samples.

I now want to try and compare them in a pair-wise manner (if it makes sense at all). I am not sure if I should use the scaling SES or the normalization parameter for the bamCompare tool. But would this workflow make sense:

1. Use the scaling factors from multiBamSummary as input for bamCoverage to scale each of the bam files

1. The bw files created in 1. I can then compare using bamCompare

==> Would this be a sound comparison of my samples?

Is there anything else I should pay attention to, when running this analysis here?

thanks

Assa