Doubt in using Homer: annotatePeaks.pl in TSS mode

Question

[k.balaji93]

So I am using HOMER to get differentially expressed peaks across 2 treatment groups, with 2 replicates each (say t1_1, t1_2 and t2_1, t2_2). I made tag directories for each of the samples separately:

makeTagDirectory t1_1 t1_1.bam -sspe -unique

Then I found peaks for each of the samples separately:

findPeaks t1_1 -style histone -o auto

I annotated the peaks with

annotatePeaks.pl tss mm10 -raw -d t1_1 t1_2 t2_1 t2_2 > raw_counts.txt

I sent the raw_counts.txt to DESEq2 for differential analysis. My Questions are:

1. Is this the right method for analyzing samples with replicates?
2. What exactly is the tss method?
3. How is it different from annotating my peaks separately and merging them with HOMER's mergePeaks?

[ejm32]

1. This could be thr right way, could not, I think it depends on the question you are trying to answer.

2. The tss method will count tags around the TSSs of genes not peaks. If you want to count tags within peaks you need to supply the peak file annotatePeaks.pl. annotatePeaks.pl <peak/BED file> <genome> [options] > <output file> http://homer.ucsd.edu/homer/ngs/quantification.html

3. mergepeaks will combine overlaping intervals. I would suggest you mergepeaks before counting tags. This way all peaks will have tag counts for all conditions/reps. Another option is to merge the bam files and then call peaks and then count tags on that dataset. This is sort of a rough approximation (very rough) of the IDR https://sites.google.com/site/anshulkundaje/projects/idr