Conditional false discovery rates in genetic association studies of rare diseases and disease subtypes
Speaker: Chris Wallace, Senior Research Fellow, MRC Biostatistics Unit, Cambridge University, UK.
This biostatistics seminar is jointly organised with the Sven Furberg Seminars in Bioinformatics and Statistical Genomics. At the end of the seminar simple food and refreshments will be served.
Genomewide association studies (GWAS) have identified thousands of variants associated with altered risk of human disease. The GWAS design enables testing a large multiplicity of hypotheses, but therefore also requires large numbers of samples. This has prevented its application to less common diseases. Additionally, while the goals of stratified medicine would be aided by a genetic understanding of disease subsets, GWAS design encourages a "lumping" rather than "splitting" approach. In this talk, I argue that the phylogeny of human diseases can be exploited to increase GWAS power in smaller samples by application of the the conditional false discovery rate (cFDR) which enables leveraging of information from related diseases. I will describe its estimation, and use examples from immune-related diseases to demonstrate increased genetic discovery. I will also describe a 2D GWAS approach for identifying aetiologically-relevant disease subtypes, and show how the cFDR can be used to identify subtype-distinguishing disease variants by leveraging overall case-control comparisons.