search

brentp / slivar

genetic variant expressions, annotation, and filtering for great good.
MIT License
252 stars 23 forks source link

Lower DP threshold for chrX variants in males #148

Closed williamrowell closed 1 year ago

williamrowell commented 1 year ago

Same justification as lowering the GQ threshold. Closes #147

brentp commented 1 year ago

Are you sure about using 3? This means that 1 of 8 variants would appear hom_ref in a het sample if all depths were exactly 3.

williamrowell commented 1 year ago

You are right, 3 does seem too low, but I'm not sure I follow about variants appearing hom_ref. For chrX variants in males outside of PAR, a few reads should be enough to indicate a variant is high quality.

brentp commented 1 year ago

Ah, right. Can you test on a few samples and verify the number of variants doesn't balloon? I'm hesitant to lower the threshold so far. Since most sequencing will be 30X illumina, this might be suprising.

williamrowell commented 1 year ago

I will, but I'll have to put it on the back burner for a bit.

brentp commented 1 year ago

No worries. how about defaulting it to 6? Then you can set config.min_male_X_dp = 3 once in a file that you send to --js.

brentp commented 1 year ago

thank you!