Title: "A Fast and Ecient Approach for Genomic Selection"
Speaker: Vitara Pungpapong, Department of Statistics, Purdue University
Place: HORT 117; February 22, 2011, Tuesday, 4:30pm


The penalized orthogonal-component regression (POCRE) combines supervised dimension reduction with Bayesian inference to select variables from high-dimensional data and to estimate the corresponding effects. It sequentially constructs linear functions of markers (i.e., orthogonal components) through supervised dimension reduction such that these components are closely correlated to the phenotype. In each construction of each component, it uses an empirical Bayes thresholding method to effectively select important markers. This construction is able to group highly correlated markers and allows for collinear or nearly collinear markers in the prediction model. Furthermore, POCRE is computationally efficient. Through simulation studies and an application to maize flowering time data, we demonstrated the promising nature of POCRE for genomic selection: fast computation and accurate prediction of breeding values. Its ability to simultaneously incorporate a huge set of highly correlated genetic markers into the model promises to make POCRE a powerful tool for meeting the challenges of genomic selection using high-throughput genotyping technologies such as the second-generation sequencing technologies.

Associated Reading:
1. Meuwissen, T.H.E., Hayes, B.J., and Goddard, M.E. (2001). Prediction of total genetic value using genome-wide dense marker maps. Genetics, 157: 1819-1829.

2. Zhang, D., Lin, Y., and Zhang, M. (2009). Penalized orthogonal-components regression for large p small n data. Electronic Journal of Statistics, 3: 781-796.

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