Director, Centre for Genomic Sciences, University of Hong Kong
*Visiting Dyason Fellow
Friday 24th May
Agar Theatre, BioSciences 4 Building, The University of Melbourne
Statistical power and risk prediction under a polygenic model
Genome-wide association studies (GWAS) conduct association tests between disease and common single nucleotide polymorphisms (SNPs), in order to identify genetic variants that influence disease risk. Under a polygenic model, numerous SNPs influence disease risk, each making a small contribution to the variation in disease liability in the population. Statistical power depends on SNP effect size, so that, given assumptions on the distribution of effect sizes across SNPs, we can derive the distribution of statistical power for a GWAS. From this, we can derive the theoretical distribution of the number of significant SNPs in a GWAS, and the expected proportion of liability variance explained by these SNPs. Furthermore, GWAS data provide coefficients for combining SNP genotypes to give a predictor of disease liability in each individual; the accuracy of this predictor is closely related to the statistical power of the GWAS.
Enquiries: Andrew Siebel (email@example.com)