Capabilities and analytical challenges in utilizing mega genomics cohort in the Veterans Healthcare System

• Activity Number: 23
• Type: Topic Contributed
• Date/Time: Sunday, July 29, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Kelly Cho*+ and David Gagnon and Hongsheng Wu and Elizabeth Lawler and J. Michael Gaziano
• Companies: MAVERIC, Veterans Administration Boston Healthcare System and Department of Veterans Affairs and VA Boston Healthcare and VA Boston Healthcare and Massachusetts Area Veterans Research and Information Center, VA Boston Healthcare System
• Keywords: Million Veteran Program ; electronic medical record ; Veterans Affairs ; genomics ; high dimensional data

Abstract

VA launched the Million Veteran Program, a nationwide genomics resource, which has over 95,000 Veterans enrolled since 2011. This provides a promising opportunity to investigate the connection between VA's longitudinal EMR and genomics data. Our understanding will highly depend on the analytical approaches used to analyze mega genomic resources. Current rapid advancement in tools to collect and extract information from genomics data, such as in GWAS, microarray or proteomics and sequence data, highlights the importance in high dimensional data analysis, including variable selection, multiple testing issues, handling, storage, and computational efficiency. Traditional statistical procedures present eminent challenges in using these data, where the number of parameters p is scalably larger than number of observations n. In addition, mega genomics data present a complex relational data structure when interactions and dynamic underlying biological complexities are considered, resulting in ultra-high dimensionality. Further research in statistical accuracy and inference, model interpretability and fitting and computational efficiency and robustness will play a critical role.