A Bayesian Hierarchical Model for Analyzing Methylated RNA Immunoprecipitation Sequencing Data

• Activity Number: 81 - New Development in Epigenome-Wide Association Studies
• Type: Contributed
• Date/Time: Sunday, July 29, 2018 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Minzhe Zhang* and Qiwei Li and Yang Xie
• Companies: University of Texas Southwestern Medical Center and University of Texas Southwestern Medical Center and University of Texas Southwestern Medical Center
• Keywords: Hidden Markov Model; Bayesian inference; MeRIP-seq; RNA epigenomics

Abstract

The recently emerged technology of methylated RNA immunoprecipitation sequencing (MeRIP-Seq) sheds light on the study of RNA epigenetics, which calls for effective and robust peaking calling algorithms. We developed a hierarchical Bayesian model to detect methylation sites from MeRIP-Seq data, which includes several important characteristics. First, it models the zero-inflated and over-dispersed counts by deploying a zero-inflated negative binomial model. Second, it incorporates a hidden Markov model (HMM) to account for the neighboring dependency. Third, the Bayesian inference allows the proposed model to borrow strength in parameter estimation to improve the model stability in small sample size setting. A Markov chain Monte Carlo (MCMC) algorithm was proposed to globally infer the model parameters. In simulation studies, the proposed method outperformed the commonly used method exomePeak, especially when the signal in the data was relatively weak or an excess of zeros were present in count count data. In real MeRIP-Seq data analysis, the proposed method identified methylation sites that were more consistent with biological knowledge, and had better spatial resolution.