B1338
Title: Integrative association analysis of multiple heterogeneous data sources
Authors: Irina Gaynanova - Texas A and M University (United States) [presenting]
Gen Li - Columbia University (United States)
Abstract: The growth of data collection and data sharing led to increased availability of multiple types of data collected on the same set of objects. As an example, RNASeq, miRNA expression and methylation data for the same tumor samples are publicly available through the Cancer Genome Atlas (TCGA). Due to the scale of the data, as well as its heterogeneity, it is typical to analyze each data type separately. We use penalized risk minimization framework as a building block for integrative association analysis of multiple heterogeneous data sources. By learning the sparse representation of underlying matrix decompositions, we are able to identify the patterns that are common across the data sources as well as source-specific patterns.
Journal