Calculating gene/exon coverage

Question

[Joshi]

Hi,

I've now read a number of posts on ways to calculate depth of coverage (for a gene, exon, etc..)

What I'm not clear about is - whether it's appropriate to count coverage with filters in place (i.e. gatk's depthOfCoverage uses a number of filters by default e.g. excluding duplicate reads, vendor quality failed reads, etc..)

If the objective of the exercise is to assess how well a particular sequencing technology is able to cover a region .. does It make sense to retain all reads? or filter out reads that meet criteria such as those applied by GATK?

Thanks, Himanshu

Answer 1

[Matteo Schiavinato]

Filters based on mapping quality, sequencing quality, or duplication are always a good idea. A read that shows one of these three problems is not reliable: on one hand you see it mapped on one particular exon, on the other you have no idea whether that is the actual biological location where it comes from.

When you do coverage analysis, it is really important that you select carefully which reads you trust, because you then use them to infer biological properties. They must then have a biological meaning.