Western Science is pleased to announce a strong initiative for development in the Science of Information with a commitment to ongoing hires in this broad area. The theme builds on the excellence in the mathematical, statistical and computer sciences at Western and complements other areas of excellence at the university: Medicine, Social Sciences, Humanities, Health Sciences, Information and Media Studies, Business and Law. This year, we seek applications for the following four related positions:
Colloquiu(Dr. Shelley Bull, WSC 248) (Thursday, November 27, 2014) Time: 3:30am-4:30am and Room: WSC 248 Speaker: Dr. Shelley Bull (Senior Investigator, Lunenfeld-Tanenbaum Research Institute of Mount Sinai Hospital, and Professor of Biostatistics, Dalla Lana School of Public Health, University of Toronto) Title: Multi-Phase and Multi-Stage Designs for Genetic Association Analysis
Statistically efficient study designs constitute an important strategy to improve cost-feasibility as we move from the GWAS* era of common genetic variants into the NGS* era of whole exome and whole genome sequencing of low-frequency and rare variants. Selection of informative individuals based on measured characteristics in combination with appropriate statistical analysis can yield cost-efficiency in population studies. Features such as genetic variation and distributional balance, and genetic model specification are key to statistical efficiency, while the sampling design must be taken into account to ensure valid analysis of the acquired data.
In a multi-phase design, Phase 1 would consist of well-phenotyped individuals, with comprehensive GWAS genotypic data for common variants. Then in Phase 2 an informative subset of these individuals would be selected for expensive in-depth characterization by fine-mapping to improve genomic resolution, for example, by dense sequencing. We consider two examples. In the first example, we investigate alternative sampling designs for selection of phase 2 subjects within strata defined by a common variant genotype available in phase 1, and use an estimating equations method to jointly analyse data from both phases. In a second example, we consider a Bayesian approach to fine-mapping study design that similarly incorporates stratification according to a promising GWAS association in the same region. In phase 2 however, improved cost-efficiency can be obtained when this fine-mapping phase incorporates a ...