bogdanlab/fizi

Check for monomorphic sites

Opened this issue · 0 comments

Currently FIZI takes as input the reference genotype data in PLINK format to compute LD matrices. There are currently no checks to test whether any variant is monomorphic or not, which breaks downstream inference.

Typically users handle this upstream by performing a MAF filter on the genotype data, but when all individuals are heterozygous, this results in MAF = 0.5, but no variation. It would be nice at a minimum to have better error handling and logging of the problem during runtime, and nicer to have seamless filtering on the fly w/o introducing extra overhead.