QC integration to metapipeline-DNA

[sorelfitzgibbon]

Here is my proposal on integrating sample level QC within metapipeline-DNA. Making some decisions will help in finalizing the SQC pipelines.

1. SAMTools stats is run directly after recalibrate BAM
   • for samples with multiple libraries
     • SAMTools stats is additionally run on each library separately.
     • insert size distributions are extracted and compared. Warnings/Errors given if libraries are too similar (indicating possible mislabeling)
     • SAMTools stats is additionally run on each readgroup separately (if > 1), or this could be triggered by failing the below QC cutoffs
   • cutoffs are set for warnings or errors based on:
     • % properly mapped and paired
     • % duplicated
     • % MQ0 (? - depends on reference)
   • Read length extracted for possible later input to CollectWgsMetrics.
2. For samples that pass above
   • start all other selected pipelines
   • additionally start FastQC (on BAM or FASTQ if available) and (mosdepth or CollectWgsMetrics)
     • note, these have much longer runtimes than SAMtools stats but take very few cpus.
     • depth calculation just at sample level
       • Consider doing depth calculations for only a few chromosomes (including Y for sex check)
     • FastQC at readgoup level to be consistent regardless of BAM or FASTQ input

Consideration is needed on how to handle libraries with thousands of readgroups.