[pomoxis] Reticulatus-pomoxis tracking issue #42
Description
Reticulatus ships with a particular version of pomoxis to provide a hacked version of the subsample_bam that runs most fast. This is a tracking issue to keep tabs on it.
- Make it work properly (recently experiments have shown it doesnt actually work which is not ideal)
- Add some randomness to the track-filling approach by keeping a list of N reads that start within the end of the emptiest track
- Raise an error if the coverage subsetting fails to yield coverage +/- some epsilon from the target