peak summit in macs not the highest point

Question

[blur]

Hi,

I ran macs on my files and took out the peak summits for further analysis. But then I opened my files in IGV, and the position of the summit does not look like the highest point in the peak.

I ran the analysis again to make sure it was not some error I did, but got the same summits. Anyone have any idea what I did wrong?

macs2 callpeak -t IP.bam -c input.bam -f BAM -g 200000000 --bw 100 --keep-dup all --nomodel -n  results -B -q 0.01

Thanks

[Gjain]

Can you please show an example of such peak (Preferably an IGV screenshot) where the summit is not the highest point?

[ATpoint]

Please also show what my files are and how these were created.

[blur]

my files are bam files, created by: alignment to reference using bwa (creating sam files) Then by samtools view -bS turned into bam files then sorted using samtools sort

[ATpoint]

I see, this is not representative in terms of visual inspection since macs2 will extend reads by the estimated fragment length. You should generate bigwig files with reads extended to fragment length, e.g. with bamCoverage from deeptools to get an idea where macs2 locates the summit.

[blur]

Is the summit file not actually a list of peak summits? or is the IGV view not accurate?

[ATpoint]

The summit file are the summits based on the pipeup that macs performs, and this comes after extending reads to fragment size. IGV based on bams is a proxy but not fully accurate for summit detection because it shows the read length that was used. If you sequenced 1x50bp for example then your reads stop after 50bp even though the true fragments are probably like 150-200bp long, and macs2 takes that into account by first estimating fragment length and then extending reads accordingly.

[blur]

Thank you for this thorough reply! I will do as you suggested :)