Using GATK MarkDuplicates for targeted sequencing data

Question

[asalimih]

Hello,
Following the gatk best practices (link), I'm using MarkDuplicates tool in my somatic variant calling pipeline. when I run the pipeline on a targeted sequenced sample (600 genes by illumina) the duplication rate is really high: In this case is it safe to use MarkDuplicates or I'm loosing a lot of informative reads which effect the certainty of the called somatic variants.

Answer 1

[ATpoint]

It seems a general consensus that you do not remove duplivates in targeted assays like this since duplication is expected by design and removing them throws away excessively much information. After all, if this is one or few amplicons per gene then you basically have only one or few unique fragment per gene or at least per haplotype so there is not really the chance to meaningfully deduplicate.