Calling peaks with macs2 if not all replicates have a corresponding input-file

Question

[jklopp]

Dear all,

I am currently trying to call peaks of a ChIP-Seq experiment using MACS2.

Not every ChIP-Seq replicate has its own corresponding input file. I was thinking of using macs2 with multiple treatment-files (tfiles) and control-files( cfiles) (even if the number of tfiles and cfiles are not the same. For example: 3 tfiles, 2 cfiles or 5 tfiles, 1 cfile) like this :

macs2 callpeak -t tfile1.bam tfile2.bam tfile3.bam -c -cfile1.bam cfile2.bam

Does macs handle this difference in file count via some kind of normalization automatically or do I have to change my set-up in a way for macs2 to work properly?

I really appreciate your thoughts on the topic!

Thank you in advance

[James Ashmore]

MACS2 will scale the number of reads in your treatment sample down to the number in your control sample. That means you can provide any number/combination of treatment and control samples and MACS2 will scale them accordingly.

[GouthamAtla]

Its need not be scaling treatment to control always. It scales the sample with large number of reads to sample with smaller number of reads.

[GouthamAtla]

I would pool all control samples and call peaks on each treatment sample independently.

macs2 callpeak -t tfile1.bam -c ctrl_pooled.bam

[me.sr1510]

How do u pool all the control sample? can it be done by samtools merge? what will happen if we merge both control and treated samples?

[jklopp]

Thank you all for you answers!