View Single Post
Old 05-14-2015, 10:08 AM   #24
Location: California

Join Date: Oct 2013
Posts: 23

The mapping quality threshold is hard-coded to only exclude unmapped or ambiguously mapped reads, see here:

Just change the -Q value to another integer if you'd like to try it yourself. However, I recall reading (can't find the reference at the moment) that keeping low-MAPQ reads did not harm copy number estimation and may have improved it.

In any case, in CNVkit the script can be used to directly exclude poorly mappable genomic regions. This is done already for hg19 in the bundled file access-5k-mappable.hg19.bed.
etal is offline   Reply With Quote