New Sparse Canonic Correlation Analysis for Construction of Co-Association Networks with Ngs Data by Cloud Computing

• Activity Number: 596
• Type: Topic Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Momiao Xiong*+ and Jin Yu
• Companies: University of Texas Health Science Center at Houston and University of Texas School of Public Health
• Keywords: next geenration sequencing ; cloud computing ; sparse ; canonical correlation

Abstract

We propose to use canonical correlation to measure the relationship between two genomic regions or genes. Then, we propose a concept of a co-association network to characterize dependence among multiple genomic regions and genes. We develop sparse canonical correlation methods for construction of genome-wide co-association networks which will be applied to studying evolution of the genome-wide co-association networks and differentiation of populations. Since the number of variants across the genome for 1,000 individuals may reach a dozens of millions of SNPs, the demands caused by genome-wide construction of co-association networks are raising great barriers in evolution studies of genome-wide co-association networks with next-generation sequencing data. To overcome these limitations, we resort cloud computations. The propose algorithms for construction and evolution studies of genome-wide co-association networks are applied to the sequencing data of 1000 genome, which is analyzed in Amazon cloud computers. In this talk, we will present the fascinating results that recover many features in evolution of population genomics and discuss how to manage and analyze data in clouds.