Filter genotypes based on other genotypes


#1

One of the QC steps I’ve started doing is that I now exclude indel LOFs from my analysis if they’re within 10bp of another indel in the same sample. I would love a way to do this filter step in Hail. I don’t think I currently can.

Edit: realized this was a genotype issue and not a variant issue. This would be like a filtergenotypes call, since I need to access sample information as well as variant information, but I also need to be able to compare genotypes to one another instead of strictly filtering based on one call’s properties.


#3

Hi kyle,

Just to understand the use case, you want to filter out indel LOF calls that are within 10bp of another indel call in that sample, yes? I had a way to do the variant level filter, but this is harder. Let me think about it.


#4

Hi Cotton,

Yes, that’s right. When we were talking about it before I neglected to specify the bit about it being in the same sample. I realized my error once I looked into how aggregating by intervals would work with this. It’s a genotype-level filter rather than a variant-level filter (although so many of these questionable calls are unique that it’s almost the same…).