Options for fq2bam/markdup to only handle optical duplicates and support multi-lane samples

Healthcare & Life Sciences Parabricks
1

I would love to see parabricks implementations of the following features currently present in GATK4’s MarkDuplicates

  1. The ability to direct markdups to only mark optical duplicates – I have non-PCR libraries in which any pcr duplicates are definitionally false-positives but I’d still like to detect optical duplicates. GATK4 MarkDuplicates provides the following useful flags:

–TAGGING_POLICY, to set the DT field for none, optical, or all duplicates, useful for downstream analysis. The default behavior of MarkDuplicates is to not write the tag field, not sure how Parabricks markdup behaves

--REMOVE_DUPLICATES, to exclude all duplicates from the output BAM file

--REMOVE_SEQUENCING_DUPLICATES, to exclude specifically the optical duplicates from the set

2. The ability to feed multiple bam files into the markdups step and produce a single duplicate-marked output bam, as is currently possible in GATK MarkDuplicates. This is useful for handling samples that have been sequenced across multiple lanes – right now there’s no good way to handle reconciliation of the duplicates between lanes without merging the FastQ files before alignment, particularly since the –no-markdups flag in fq2bam doesn’t seem to work

2

Through closer reading of the FQ2bam documentation I realized that multi-sample inputs are actually supported already, which is great and is saving me a lot of overhead on my projects. The duplicate tagging/removing options would still be really useful additions