GWAS with more covariates than observations

Hi there,

I ran a GWAS in Hail and found that some of the SNPs with sample size smaller than the number of covariates actually have non-empty BETA and SE in the output summary statistics. However, the P-value is always NA in this case.

I wonder if this has something to do with Hail’s algorithm in fitting an OLS model with more predictors than observations? Thanks!