BayesSUR: An R Package for High-Dimensional Multivariate Bayesian Variable and Covariance Selection in Linear Regression

In molecular biology, advances in high-throughput technologies have made it possible to study complex multivariate phenotypes and their simultaneous associations with high-dimensional genomic other omics data, a problem that can be studied multi-response regression, where the response variables are potentially highly correlated. To this purpose, we recently introduced several Bayesian variable covariance selection models, e.g., estimation methods for sparse seemingly unrelated regression selection. Several priors been implemented context, particular hotspot detection prior latent inclusion indicators, which results between predictors multiple phenotypes. We also propose an alternative, uses Markov random field (MRF) incorporating knowledge about dependence structure of indicators. Inference (SUR) by chain Monte Carlo is computationally feasible factorization matrix amongst variables. paper present BayesSUR, R package, allows user easily specify run range different SUR C++ computational efficiency. The package specification models modular way, chooses separately. demonstrate performance spike-and-slab MRF on synthetic real data sets representing eQTL or mQTL studies vitro anti-cancer drug screening as examples typical applications.