gravatar for i.sudbery

12 minutes ago by

Sheffield, UK

I don't know exactly how Qualimap measure duplication, but a quick look at the documentation strongly suggets to me that it measures the single-end duplication rate, where as Picard will remove reads on the basis of the paired-end duplication rate. In single-end duplication, for a read to be a duplicate, it need only share its alignment start position with another read, where as for pair-end duplication a read must share both its start postion, and the position of its mate with another read in order to be called a duplicate.

Where you have paired-end data you almost always care about paired duplications, not single-end duplicates.

Source link