American Statistical Association New York City Metropolitan Area Chapter Memorial Sloan-Kettering Cancer Center Biostatistics Seminar

Recent technological progress has led to rapid identification and sequencing of large numbers of genetic variants. Studying the associations between these variants and a particular disease is of great importance to epidemiologists in their quest to decipher the disease etiology. Hierarchical modeling is a technique which has been shown to provide more accurate and stable estimates of individual variants, by incorporating external information through the higher levels of the multilevel model. This talk presents recent research on the application and implementation of hierarchical modeling regression to handle these issues using pseudo-likelihood and Gibbs sampling methods. A real data set from a melanoma case-control study is used to illustrate the methods.