Duplicated Reads for Somatic Mutation Detection

[LiLin]

Hello, everyone,

There is one question about somatic mutation detection, is there anyone could help me?

Duplicated reads may be marked by Picard in bam-files, and most software don't care this tag. When we calculate CNVs, we shouldn't use duplicated reads, How about Somatic SNVs or InDels?

[Bukowski]

No, you need to ignore them just as you would for all variant-calling applications. If they're duplicates then they're PCR or optical duplicates and potential sources of bias. To eliminate the bias remove them or mark them for downstream variant calling applications. About the only time this isn't done routinely is for RNA-Seq expression studies.