Calculating fold 80 base penalty by hand

Question

[Seesea]

Hello,

Is there a simple formula for calculating fold 80 base penalty without using Picard metrics? For example using knowledge about number of sequenced bases, mean coverage and standard deviation of mean coverage of WES experiment?

Answer 1

[Istvan Albert]

In my opinion, the main problem is that the coverage across targets typically strongly diverges from a mathematical model.

Hence I think mean coverage and standard deviation are probably insufficient to figure out the penalty.

But the post made me think about ways to compute this easily and I gave it a go, I am hoping I am not completely off here:

The fold 80 base penalty is defined as the fold change of non-zero read coverage needed to bring 80% of the ROI bases to the observed mean coverage.

What I think you could do, is compute the coverage for each target, sort that list by coverage, and find the coverage of the 80% percentile. The ratio of that target's value from the desired coverage will be the fold 80 penalty.

For example if the 80% percentile's target is 10x and you wanted 20x then your fold penalty is 20/10 = 2