Detecting Dynamic Long-Range Chromatin Interactions from Multiple Cell Types Hi-C Data Using a Bayesian Hierarchical Hidden Markov Random Field Model

• Activity Number: 401
• Type: Invited
• Date/Time: Tuesday, August 2, 2016 : 2:00 PM to 3:50 PM
• Sponsor: Committee of Representatives to AAAS
• Author(s): Zheng Xu and Guosheng Zhang and Anthony Schmidt and Bing Ren and Ming Hu and Yun Li*
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and University of California at San Diego and University of California at San Diego and New York University and The University of North Carolina at Chapel Hill
• Keywords: chromatin interaction ; Hi-C data ; Bayesian hierarchical hidden Markov random field

Abstract

The advent and rapid development of chromosome conformation capture (3C)-based technologies, in particular its genome-wide extension Hi-C technology, enable a genome-wide view of chromosome spatial organization. The constantly accumulating Hi-C datasets provide rich information for studying three-dimensional (3D) genome organization across multiple cell differentiation stages. However, the statistical models and computational tools for detecting dynamic long-range chromatin interactions (i.e., peaks) are still lacking. Limited sequencing depth within each Hi-C dataset, as well as heterogeneity among different Hi-C datasets, pose great challenges for downstream data analysis and data interpretation. To fill in these gaps, we here develop a Bayesian hierarchical hidden Markov random field (HHMRF)-based model to detect dynamic long-range chromatin interactions from Hi-C data collected from different cell types. Our method can effectively borrow information from multiple Hi-C datasets, therefore achieve higher robustness and enhanced statistical power. We applied HHMRF to analyze Hi-C datasets on human embryonic stem cells (H1 ESC) and four H1 derived cells, and identified a large numb