Are HOMER aggregate proflles normalized to the read depth of each ChIP-seq sample? [still open]

Question

[biostart]

Hello,

I am comparing HOMER generated aggregate profiles of my ChIP-seq for two cell conditions. The shapes of the aggregate profiles look the same, but they are shifted vertically. Should I assume that HOMER is doing the internal normalization of each BED file to the read depth? Is it safe to compare the heights of the aggregate profiles generated by HOMER in two cell conditions?

Thank you!

Answer 1

[bede.portz]

Provide some more detail, please. Are you using annotatePeaks -hist? If so, the default is to normalized the read files to 10,000,000 reads. See the link below and scroll to the bottom under the heading "advanced options".

I hope this helps.

http://homer.salk.edu/homer/ngs/annotation.html

[biostart]

Yes, using annotatePeaks -hist. I did not do any prior normalization. The question is, whether HOMER is doing proper normalization. I have several replicates for each of the two cell conditions. Calculating HOMER profiles for each replicate, then average all replicates for one cell condition, and separately all replicates for another cell condition. As a result I have two HOMER profiles for two cell conditions. They are shifted vertically with respect to each other. When I am increasing the window size (-size 2000; -size 20000, etc), I still get a similar situation that the two profiles are shifter vertically with respect to each other. I am having hard time to convince my collaborators (and myself) that the effect is real, and not some HOMER artifact. Any ideas?