This is a very related problem with MACS2. I am doing peak calling of my 'chirp' seq data, using SAM files, treatment and input contain tag length of 126bp. I noticed the classic 'Since the d (20) calculated from paired-peaks are smaller than 2*tag length, it may be influenced by unknown sequencing problem!' warning upon peak calling. Although it skipped building the model, it did move ahead with calling the peaks and generated bed/xls/bedgraph files.
Then, I tried the --no model --ext size 200 peak calling command, although the command goes through without any error/warning, the output files obtained from this are 'blank'.
I also used spp to determine my fragment size and it gave outputs as 180 (treatment) and 173 (control).
At this point of my analysis, I am mainly interested in my bed and xls file outputs, which I am using to do peak correlation after. I was wondering if there is an explanation to why I would get blank outputs with no-model and if it was ok to go ahead with the bed/xls files generated from my run with the warning.
Also, since the original error was that d<2tag length, I was wondering if I should use a fragment length of higher than 2tag length?
Any suggestions/ discussion is greatly appreciated.