This repository contains source code to reproduce analyses in "Valid inference for machine learning-assisted genome-wide association studies".
The official software package is in POP-TOOLS GitHub repo.
./simulation/Fun.R: Functions for simulation and simple implementation of POP-GWAS in R./simulation/qt.R: simulation for quantitative phenotype./simulation/bt.R: simulation for binary phenotype./simulation/imputation_r_FPR.R: type-I error simulation for varying imputation correlation./simulation/imputation_r_power.R: power simulation for varying imputation correlation./simulation/vary_ratio.R: simulation for varying sample size of unlabeled data
./real_data/1_softimpute.R: phenotype imputation using softimpute./real_data/2_regenie.sh: run GWAS using regenie./real_data/3_popGWAS.sh: apply POP-GWAS to summary statistics./real_data/4.1_post-GWAS.sh: estimating heritability and genetic correlation using LD score regression./real_data/4.2_coloc.R: colocalization analysis
Feel free to reach out to Jiacheng at jmiao24@wisc.edu for questions.