Integrated analysis of diverse genomic data

Abstract

The increasing growth of high throughput genome-wide assays, such as next generation sequencing, is enabling the simultaneous measurement of several genomic features in the same biological samples. Recently many studies consider integrating the available data aiming for a more comprehensive understanding of the genome. Along these lines, we consider publicly available diverse data derived from The Cancer Genome Atlas database and for the same samples, to explore the biological merits of integration comparatively to single data analysis. A partition model is presented to detect interactions across data sets. Both simulated and empirical data examples demonstrated our method’s ability to detect highly correlated data groups across platforms and provided key insights into previously defined gene expression subtypes.