mzekavat
commented
8 years ago sample mmskat group test:
/humgen/gsa-hphome1/sek/akihiro/Software/EPACTS-3.2.6/bin/epacts group --groupf ${GROUP}.txt \ --vcf ${VCF_dir}/FinEst.WGS.QCed_Lipids.chr1.vcf.gz \ --ped ${Phen_dir}/EstoniaOnlyLipids.ped --max-maf 0.01 \ --kin ${Kin_dir}/FinEstKinship.kinf --sepchr --pheno ${PHENO} \ --cov AGE --cov FAST10 --cov SEX \ --cov LCSET.7048 --cov LCSET.7049 --cov LCSET.7123 --cov LCSET.7125 --cov LCSET.7130 --cov LCSET.7131 --cov LCSET.7132 --cov LCSET.7148 --cov LCSET.7162 --cov LCSET.7174 --cov LCSET.7175 --cov LCSET.7176 --cov LCSET.7257 --cov LCSET.7263 --cov LCSET.7308 --cov LCSET.7331 --cov LCSET.7334 --cov LCSET.7339 --cov LCSET.7340 --cov LCSET.7341 --cov LCSET.7353 --cov LCSET.7354 --cov LCSET.7355 --cov LCSET.7356 --cov LCSET.7357 --cov LCSET.7419 --cov LCSET.7420 --cov LCSET.7473 --cov LCSET.7494 \ --test mmskat --out ${OUT}.ESTonly --run 4
tpoterba
Hello, It would be really great to have the following tests in Hail, as detailed on EPACTS: http://genome.sph.umich.edu/wiki/EPACTS#Single_Variant_Test
Most important for immediate analyses (with ~10K individuals WGS) are q.emmax, emmaxCMC, and mmskat, which all use mixed models (with kinship matrices or GRMs).
Furthermore, one step beyond running analysis is doing conditional analysis. Right now, in EPACTS, doing conditional analysis requires adding separate columns in the .ped file corresponding to GTs for each variant you'd like to condition on. Ideally, we'd be able to just list the variants (maybe in tab-delimited format with Chr, Pos, Ref, Alt.
quantitative traits of interest: q.emmax mmskat
binary traits of interest: b.score (or b.wald) b.collapse emmaxCMC
Thanks again!