BhGLM is an R package. This package provides functions for setting up and fitting various Bayesian hierarchical models (generalized linear models (GLMs), Cox survival models, ordered logistic or probit regressions, truncated regressions, and conditional logistic models), for numerically and graphically summarizing the fitted models, for evaluating the predictive performance, and for hypothesis testing using Wald-type tests and score tests. Four types of priors on the coefficients can be used: double-exponential, Student-t, mixture double-exponential, and mixture Student-t. The priors can incorporate the group structure and networks of predictors into the model. The methods can be used to analyze not only general data but also large-scale genomic data (i.e., detecting disease-associated genes or variants and predicting phenotypes), and can deal with various types of continuous, discrete and censored phenotypes in population-based genetic studies as well as family-based (matched) case-control studies and extreme phenotype sampling designs.
- Nengjun Yi, Ph.D.
Sir David Cox Endowed Professor
Section on Statistical Genetics
Department of Biostatistics
Ryals Public Health Bldg, 317F
University of Alabama at Birmingham
Birmingham, AL 35294
Phone: (205) 934-4924
Fax: (205) 975-2540