This repository contains an R package for creating polygenic risk scores from rare variants using RICE. Specifically, this package creates a polygenic risk score using RICE-RV (as denoted in the manuscript) with significant rare variant sets and their burden scores extracted from an AGDS file. A detailed example of this is provided in the vignette. Note, if an individual has burden scores from a source other than an AGDS file, a rare variant PRS following the RICE-RV steps can still be created by starting at the Penalized Regression step in the vignette.
RICE can be viewed as an extension to STAARpipeline. We suggest users familiarized with STAARpipeline and it's tutorial before constructing rare variant PRSs using RICE.
R (recommended version >= 4.0.0)
RICE imports R packages Rcpp, SCANG, dplyr, SeqArray, SeqVarTools, GenomicFeatures, TxDb.Hsapiens.UCSC.hg38.knownGene, GMMAT, GENESIS, Matrix, methods, caret, glmnet, SuperLearner, STAAR, and STAARpipeline. These dependencies should be installed before installing RICE.
library(devtools)
devtools::install_github("jwilliams10/RICE")Please see the RICE user manual for detailed usage of RICE package.
An uncompiled vignette and a compiled vignette is available that showcases how to create a rare variant PRS with part of chromosome 22 for 1000 Genome data. Data for the example in the vignette is downloadable through Harvard Dataverse.
Please direct any problems or questions to Jacob Williams jacob.williams@nih.gov.
If you use RICE in your work, please cite:
Jacob Williams*, Tony Chen, Xing Hua, Wendy Wong, Kai Yu, Peter Kraft, Xihao Li*, & Haoyu Zhang*. (2024). Integrating Common and Rare Variants Improves Polygenic Risk Prediction Across Diverse Populations. medRxiv. DOI: 2024.11.05.24316779
This software is licensed under GPLv3.
