Comparison of bamCoverage log2 ratios in deeptools

Question

[ATpoint]

Is anyone experienced in using bamCompare from deeptools? I ran it to compare a ChIP sample to its input, using SES normalization, outputting the log2 ratio. If I do the same on a replicate with its own input, can I then directly compare the log2 ratios for the two outputs in a certain region (or the genome), e.g. for a sample-to-sample correlation plot?

bamCompare -e 200 -bs 1 --scaleFactorsMethod SES -o out_log2ratio.bigwig -b1 sample.bam -b2 input.bam

Answer 1

[Devon Ryan]

Yes, you can directly compare the SES normalized replicates. Keep in mind that you're then looking at not only the difference in signal distribution between the replicates, but also how well those replicates were enriched relative to their inputs. In general, SES makes sharper marks more distinct (this is good), but if you have either a poorer enrichment in one sample or a weird input sample then you might get a bit poorer comparison than otherwise.