Variant depth cut off - justification

Question

[amy__]

Hello,

I am trying to find some studies or reports that suggest exactly what type of variant depth we should be using as a cut off in WES or general variant calling - I have seen the numbers 5, 10 and 30 flying around but I want to be able to justify my cut off whilst also not cutting off potentially real variants.

Does anyone know of any good papers or studies? I have had a google but can't find anything great. These are not trios.

Thanks! Amy

Answer 1

[dthorbur]

I think picking a threshold needs to be made in relation to the data. If you have a target coverage threshold of 100x, then 5x would be a very low threshold, and 30x would be more reasonable.

I would plot your coverage, so you can see how much data you'd be losing and make an informed decision rather than picking an arbitrary value from the literature.