abnerzyx/pv: Prism Vote: A stratified statistical framework for individual disease risk prediction

The Prism Vote framework is in essence a Bayesian hierarchical probability model, in which an individual’s disease risk is integrated from subpopulation-specific disease estimates and the prior probabilities of a subject’s subpopulation identity. Using PLINK, Principal Component Analysis (PCA) is adopted to identify subpopulations, Logistic Regression (LR) analysis is adopted for case-control Genome-Wide Association Studies (GWAS). Feature selection is performed using GWAS P-values with user-specified cutoff or using user-specified candidate SNPs. Four prediction models are built on selected SNPs. Two are LR and Polygenic Risk Score (PRS)-based LR. The other two are LR and PRS-based LR under the Prism Vote framework. Covariates can be included in GWAS analysis and predictive modeling.

install.packages("remotes")
remotes::install_github("abnerzyx/pv")