American Statistical Association New York City Metropolitan Area Chapter Mailman School of Public Health Columbia University Department of Biostatistics Colloquium

Modern association studies of complex traits, such as GWAS or sequencing studies, demand statistical tools that are capable of detecting small-effect variants, model complex interaction effects, and have convincing speed performance. In this work, we introduce a similarity-based regression method to perform marker-set analysis. The method uses genetic similarity to aggregate information from multiple polymorphic sites (e.g., SNPs or a mixture of different polymorphisms), and regresses trait similarities for pairs of unrelated individuals on their genetic similarities to access the gene-trait association. The association is detected using a score test whose limiting distribution is derived. The proposed method can account for covariates, has the capacity to model both main and interaction effects, and is computationally efficient. We also show that the gene-trait similarity regression does not require phase sequence and that it explicitly models the non-additive effects among markers. These features makes it an ideal tool for evaluating association between phenotype and marker sets defined by haplotypes, genes or pathway in whole-genome analysis.

Dr. Tzeng received her Ph.D. in Statistics from Carnegie Mellon University in 2003 under the direction of Dr. Kathryn Roeder. She is currently Assistant Professor in Statistics at North Carolina State University, which she joined in the fall of 2003. Her research interests combine the fields of statistics and genetics, and she focuses on developing statistical methods that can facilitate genetic epidemiologic research on human complex diseases. Some of her current research projects include statistical modeling of multimarker/haplotype association for genome-wide and candidate-gene studies, gene-based and pathway-based analysis for pharmacogenetics, and SNP genotyping error and quality control.