Title: "New Statistical Methods for Simultaneous Genome-Wide Association Studies"
Speaker: Min Zhang, Department of Statistics, Purdue University
Place: LILY G126; February 17, 2009, Tuesday, 4:30pm

Abstract
Current genome-wide association studies are conducted by genotyping an overwhelmingly large number of single nucleotide polymorphisms (SNPs) from a small number of individuals, and investigating the association between each SNP and the phenotype of interest. However, such univariate approaches ignore the complicated correlation structure between SNPs. The potential collinearity issues also challenge the classical multivariate approaches. We proposed a penalized orthogonal-components regression (POCRE) method to simultaneously model tens of thousands of SNPs. The proposed method builds sparsely loaded principle components through supervised learning, and it can be applied to both quantitative-phenotype and case-control data.


Recommended Reading
Zhang, D, Lin, Y and Zhang, M. (2008). Penalized Orthogonal-Components Regression for Large p Small n Data. < http://arxiv.org/PS_cache/arxiv/pdf/0811/0811.4167v3.pdf>

Hirschehorn, J.N., and Daly, M.J. (2005). Genome-Wide Association Studies for Common Diseases and Complex Traits. Nature Reviews Genetics, 6:95-108.



Click here for a full schedule of BIOINFORMATICS SEMINARS, past and present.