Minimal mapping score of bowtie2

Question

[dcheng1]

I'm trying to filter out low-quality alignment before calculating correlation coefficient. What's the cut-off mapping score people usually use to get rid of bad alignment for bowtie2?

[ATpoint]

I personally use MAPQ 20 for standard tasks like ATAC-seq or ChIP-seq. But don't ask why 20 and not 25, 15 or 10. I think I stumbled over 20 which was used by a reputable tool when I got started with NGS so I took it over. Unless you are investigating difficult genomic regions specifically, choice should not matter too much as long as it is > 0 to avoid multimappers.

[dcheng1]

Thanks for your help!

[colin.kern]

The ENCODE pipeline uses a MAPQ 30 cut-off