Genomic Data Integration to Improve Disease Subtype Discovery

• Activity Number: 156
• Type: Invited
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: WNAR
• Author(s): George Tseng*+ and Sunghwan Kim and Caleb Huo
• Companies: Univ of Pittsburgh and University of Pittsburgh and University of Pittsburgh
• Keywords: genomics ; integrative analysis ; meta-analysis ; disease subtype

Abstract

Many complex diseases were once thought of as a single disease but genomic studies have revealed their disease subtypes that contain different disease mechanisms, survival outcome and needs of different treatments. As high-throughput experiments become more and more affordable, many data sets of the same omic type (e.g. transcriptome) and of a related disease have been accumulated. The horizontal meta-analysis to combine multiple studies has become an important task. On the other hand, many types of omic data are now often generated in a given cohort (e.g. genotyping, transcriptome, methylation etc). The vertical integrative analysis brings even greater challenges. In this talk, we will present two genomic data integration methods for disease subtype discovery. The first method, MetaSparseKmeans, performs horizontal meta-analysis to combine multiple transcriptomic studies. The second method, integrative Phenotyping Framework (iPF), focuses on vertical integrative analysis and integrates multi-omices data of a given cohort. We will apply the methods to breast cancer microarray data, the Lung Genomic Research Consortium (LGRC) and The Cancer Genome Atlas (TCGA) multi-omics data.