A fully Bayesian latent variable model for integrative clustering analysis of multi-type omics data

• DOI: 10.1093/biostatistics/kxx017
• Publication Date: 2017-04-14T15:08:43Z
• Authors: Qianxing Mo; Ronglai Shen; Cui Guo; Marina Vannucci; Keith S Chan; Susan G Hilsenbeck
• Keywords: Omics; Identification; Epigenomics; Data type

Identification of clinically relevant tumor subtypes and omics signatures is an important task in cancer translational research for precision medicine. Large-scale genomic profiling studies such as The Cancer Genome Atlas (TCGA) Research Network have generated vast amounts genomic, transcriptomic, epigenomic, proteomic data. While these provided great resources researchers to discover driver molecular alterations, there are few computationally efficient methods tools integrative clustering analysis multi-type Therefore, the aim this article develop a fully Bayesian latent variable method (called iClusterBayes) that can jointly model data continuous discrete types identification features. Specifically, proposed uses variables capture inherent structure multiple sets achieve joint dimension reduction. As result, samples be clustered space features drive sample identified through selection. This significantly improve on existing iClusterPlus terms statistical inference computational speed. By analyzing TCGA simulated sets, we demonstrate excellent performance revealing meaningful