Statistical Methods for Genetic Analysis
Complex traits, including disease and disabilities, that vary in human and natural populations are determined by multiple genetic and environmental factors that interact with one another in complicated, often non-linear ways. Our recent work is devoted to developing new statistical tools for genetic research that are effective for analyzing highly multidimensional data and modeling complex traits in individuals and families.
Association tests are used to inspect genetic effects in populations of individuals. A particularly robust approach to detecting influential genes is transmission disequilibrium testing (TDT). We have rigorously established the foundation of TDT and have extended it to handle a wide variety of genetic and environmental effects and their interactions.
Nonparametric linkage analysis uses related individuals to discover genetic effects. We have adapted our TDT formulation to develop new tests for linkage using nuclear families which can detect complex interacting genetic and environmental effects.
Our most recent interest is the analysis of gene expression data from microarrays. We are developing an approach to cluster analysis of gene expression arrays which incorporates information about clinical characteristics and prior information about gene function. The algorithms are very efficient - the speed varies linearly with the number of genes.
We view cluster analysis as the crude first step in describing the network structure of genomic expression. Our future work will be directed toward identifying more complex network representations.
