时间：2016年1月6日(周三)上午10:00-11:00
地点：闵行校区统计楼105室
报告人：多伦多大学公共卫生学院徐竞雄博士
题目：A Novel Multiple-SNP Approach for Fine-Mapping Studies
摘要：
A fine-mapping study focuses on the investigation of association between the clustering of tightly linked genetic markers from the same chromosomal region and a complex trait of disease. Fine-mapping studies usually involve genotyping additional markers in the region of interest or alternatively imputing them using a reference genetic panel. Fine mapping studies provide opportunities to identify the causal loci underlying complex human diseases but also yield many statistical challenges arising from the complex correlations and high linkage disequilibrium (LD) between the markers.
As an illustrative example, the Kallikrein (KLK) region on chromosome 19 entails an important family of genes, clustered in a region that spans 261,558 bp, that display significant homology to each other at both the nucleotide and protein levels. The Prostate Specific Antigen (PSA) marker is the serum level of the KLK3 protein. To better understand the implication of SNPs in the KLK region and their association with prostate cancer (PCa), we performed a fine-mapping study of the entire KLK region. The study included 123 original SNPs and 925 imputed SNPs from a sample of 920 cases and 904 controls. To model the joint effect of SNPs and account for their complex dependence structure, we applied a novel approach based on Bayesian Graphical Model (BGM) and an efficient algorithm, MOSS, to perform the model search and SNP selection. We performed simulation studies to assess and compare the performances of BGM and other competing methods. The data analysis showed two genetic paths leading to PCa, one is the region close to KLK4 gene and another one is between KLK12 and KLK13.