Samtools markdup hits walltime with high depth regions

[denriquez]

Tool: Samtools markdup (v1.10)
Issue: Runs way past 8 days on a KHWGS BAMs (doesn't complete)
Cause of Issue: More than 8 million reads mapping in the chr2:32916200 – 32916600 (~400 bp) region
Samples encountering issue info: Fresh frozen SFG brain, Kapa Hyper WGS
Solution attempt:

1. Ran markdups per chromosome still hit walltime on chr2.
2. Created microbams (chr2:32911150-32921650) and ran samtools markdup v1.11 and gatk MarkDuplicates v4.1.8.0 on them. Samtools markdup ran past 3 days (didn't complete) while gatk MarkDuplicates finished under 35 minutes. The extreme case (21 million reads within the region specified above) finished within a 93 minutes.

[PedalheadPHX]

@denriquez so the new version of Samtools did not complete even on the micro-bam? If so lets push the issue to the samtools team

[denriquez]

Correct. Will do.

[PedalheadPHX]

@denriquez can you added the issue number from the samtools issues list so we can easily link to your open issue there to follow the status

[denriquez]

Github Samtools issue #1325 [samtools/samtools/issues/1325]

[PedalheadPHX]

Waiting for formal release with the fix