Deduping AFTER BQSR
Is there any justification for putting off the MarkDuplicates step until after you've run BQSR? Based on my knowledge of bioinformatics, this seems like a dangerous idea, but it's the one recommended in this samtools workflow:
Basically, they say, run
1. RealignerTargetCreator and IndelRealigner , then
2. BaseRecalibrator and PrintReads , then
I apologize for asking questions about another institute's workflow, but I feel like you'd be the folks most likely to know whether doing things in this order has any advantages. Thanks for any help you can provide.
By the way, I'm working on single-cell DNA-seq cancer data.