Statistical Methods for Integrative Genomics

• Activity Number: 8
• Type: Invited
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Joseph Beyene*+
• Companies: McMaster University
• Keywords: Data integration ; association ; prediction ; sparse methods

Abstract

There is an increasing recognition of the utility of integrating diverse data in the biological and other sciences. For instance, integrative genomics studies in which different types of genomic data have been integrated (e.g., genotype and expression data) have proven to be useful in advancing our understanding of basic biological processes and are expected to shed new light into our ability for predicting outcomes of interest with improved accuracy. However, there are several statistical and methodological challenges associated with data integration. In this talk, I will present some of the fundamental concepts and challenges involved in data integration, and will share lessons that we have learnt from attempting to integrate high-dimensional genomic data. I will present results from comprehensive simulation studies as well as illustrative real data examples. I will highlight the strengths and limitations of the various integrative approaches.