BQSR filters out reads 'not in target region' about elprep HOT 2 CLOSED

@matthdsm This is actually a "feature" in the sense that it reproduces exactly the same behavior as in GATK: When you pass the -L option to ApplyBQSR, it also produces a BAM file that filters out reads that are not in the specified regions. This also has an impact on HaplotypeCaller: When you pass the -L option to HaplotypeCaller, it will work only on the specified regions, but adds some padding around those. If we wouldn't filter out reads in our BQSR step, our HaplotypeCaller would therefore take more reads into account than the original HaplotypeCaller. (GATK's ApplyBQSR with -L option effectively cancels out the padding in GATK's HaplotypeCaller with -L option.)

If you have a better idea how we should handle this case while remaining compatible, we would be very interested to hear suggestions. :)

Pascal

from elprep.

Hi Pascal,

Thanks for the clarification.
I get the need for feature parity between Elprep and the GATK, it's a big selling point. With this explanation, I think it's handled as it should be.

Thanks for clearing up, feel free to close.
Cheers
Matthias

from elprep.