An HMRF-Based Bayesian Method for Chromatin Interaction Calling from Hi-C Data

• Activity Number: 603
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Zheng Xu*+ and Guosheng Zhang and Fulai Jin and Ming Hu and Yun Li
• Companies: University of North Carolina at Chapel Hill and University of North Carolina at Chapel Hill and University of California, San Diego and New York University and University of North Carolina at Chapel Hill
• Keywords: Bayesian Estimation ; Chromatin Interaction ; Contact Matrix ; Hidden Markov Random Filed (HMRF) ; Spatial Organization ; Peak Calling

Abstract

Advancements in chromosome conformation capture and next generation sequencing technologies are enabling genomewide investigation of dynamic chromatin interactions. Model-based methods to detect biologically meaningful chromatin interactions from massive random chromatin interactions are still lacking. To analyze the Hi-C dataset with the highest resolution to date (Jin et al 2013), Jin et al developed a one-dimensional peak calling method thresholding on pairwise p-values and read counts. We propose a Hidden Markov Random Field (HMRF) based Bayesian method to rigorously model the probability of interaction in the two-dimensional contact frequency matrix. Comparing with the one-dimensional peak calling method, our two-dimensional peak caller has the following desirable properties (1) symmetry and robustness to analysis unit; (2) quantitative estimation of peak probability; (3) flexibility and rigor in FDR control; (4) allowing the incorporation of biologically relevant priors; and (5) borrowing information from surrounding loci pairs to improve statistical power. We have shown that our method has better performance in both simulations and real data.