I wish to perform resampling by bootstrapping or a similar method so that I can obtain empirical p-values for my data. The purpose of this is to correct for non-normality while avoiding overcorrection that comes with transformation of data.
I'm currently using SKAT, which allows one to build null models that include resamples and then test the alternative models against these null models. This is much faster than permuting phenotype data and re-running the test. I'm looking for a way to do something similar in a GWAS tool for single-locus testing, such as GENESIS.
Ideally, I would like the amount of resampling to depend on how long the p-value is, since there's no need to resample 1 million times for a p-value of 0.5, or contrarily, 1000 resamples would be insufficient for a p-value of 1e-5, which would probably appear to be 0 without enough resampling.
I haven't been able to find a built-in ressampling function in GENESIS, but does this exist for GENESIS or a similar GWAS package that tests SNPs one at a time (unlike SKAT which tests for combined effect of SNPs)?