I am new to ChIP-seq analysis and I am proceeding with bamCompare to generate a .bw file from two .bam files (one ChIPed and one input). In my ChIP experiment I am investigating enrichment of a very broad histone mark, which can cover up to megabases along the genome.
To do so, I am using the following command line:
bamCompare -b1 aln_results/aln_sorted/WT_1.1_NT.srt.bam -b2 aln_results/aln_sorted/WT_1.1_NT.input.srt.bam -o IGV/WT_1.1_NT/WT_1.1_NT.scale.bw -p 3 -bl hg19/blacklist/blacklist_hg19.merged.bed --effectiveGenomeSize 2864785220 -e --ignoreDuplicates -bs 20000 --smoothLength 200000
When I visualize my output in the IGV genome browser, I notice that almost all the genomic regions show negative values, which makes me wonder a bit. I am aware of the fact that, by default, the --operation parameter is set to log2 and, therefore, negative values are expected. However, for visualization purposes, I tried to set "--operation mean" and now I got positive values in the output and in the genome browser track.
Consequently, I am wondering how appropriate is to set --operation to "mean" or any other option, rather than computing log2 ratio between ChIP and input samples.
Moreover, would it be appropriate to also set a normalization method, such as BPM, or would you suggest to use scaling factors instead?