Q: Guidance on enhancing MACS3 parameters.

[fm504]

Hi,

We are currently working with paired-end data and utilizing macs3 callpeak for peak calling. After, exploring the different options available, we read that it is highly recommended to use the --nomodel option when establishing comparisons between datasets. We've applied callpeak to four different samples without a control, as we are not analyzing ChIP or ATAC-Seq. The command used for peak calling is attached:

macs3 callpeak -t sample.bam -g hs --format BAMPE --cutoff-analysis --keep-dup all --SPMR -B --trackline -n sample

We would like to ask whether specifying --SPMR would be sufficient for robust comparisons among these samples.

Additionally, we are exploring various options for --keep-dup parameter. In particular, we are interested in accurately computing the actual coverage of the samples after duplicate removal done by macs3.

For --keep-dup = all, it could be approppiate to utilize bedtools coverage directly on the bam file, as no read have been eliminated.

bedtools coverage -a sample_peaks.narrowPeak -b sample_bwa_sorted.bam -mean > MeanCoverage${file}_dup.bedgraph

However, we are considering if it could be an appropiate alternative to use bedtools map,, to compute coverage from the bdg files generated by macs3, particularly for --keep-dup = 1 or --keep-dup= all , where some reads has been eliminated to compute the peak calling.

bedtools map -a sample_peaks.narrowPeak -b sample_treat_pileup_sorted.bdg -c 4 -o mean

Any insight or recommendation would be greatly appreciated!.