A Hidden Ising Model for ChIP-chip Data Analysis

• Activity Number: 44
• Type: Contributed
• Date/Time: Sunday, August 1, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Qianxing Mo*+ and Faming Liang
• Companies: Memorial Sloan-Kettering Cancer Center and Texas A&M University
• Keywords: ChIP-chip ; Hidden Markov Model ; Ising model ; Bayesian hierarchical model ; genomic data ; spatial model

Abstract

Hidden Markov models are widely used to model the spatial dependency of ChIP-chip data. However, parameter estimation for these models is typically either heuristic or suboptimal, leading to inconsistencies in their applications. We propose a hidden Ising model that can overcome this limitation. The data are modeled in a fully Bayesian framework, and Metropolis within Gibbs sampling algorithm is used to simulate from the posterior distribution of the model parameters. The proposed model naturally incorporates the spatial dependency of the data, and can be used to analyze data with various genomic resolutions and sample sizes. We illustrate the method using three publicly available data sets and various simulated data sets, and compare it with three Markov model-based methods. We find that our method significantly outperforms the alternative methods for the low resolution data.