HI I have an insane coverage in my bamfiles (thanks to only 12way multiplexing of a tiny genomic region). Since some programs can't handle insane coverages, I would like to downsample, that is pick say 1000 aligned reads for each position randomly (E.g. with bamtools random: Select random alignments from existing BAM file(s)) but I need to keep the mates intact. Another way of putting this: I want to achieve a maximal read-depth of 1000 for each position.
Has anyone experience with downsampling? Are there any pitfalls I might not be aware of?