HierCM: a Hierarchical Mixture Model Approach for Detecting Chromatin Interactions in Hi-C Data

• Activity Number: 527 - Contributed Poster Presentations: Section on Statistics in Genomics and Genetics
• Type: Contributed
• Date/Time: Wednesday, July 31, 2019 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Frank Shen* and Qunhua Li and Naomi S Altman
• Companies: Penn State University and Penn State University and Pennsylvania State University
• Keywords: Hi-C; Chromosome Conformation Data; Mixture Model; Peak-calling

Abstract

Hi-C chromosome conformation data may reveal novel mechanisms for gene regulation, but analysis is difficult due to high variability and numerous biases. We introduce a new probabilistic hierarchical mixture model, HierCM, for detecting chromatin interactions by integrating replicate measurements separately and incorporating the TAD-caller TopDom into our estimation method, which improves our background estimator . This improves HierCM peak-calling reproducibility when using replicates, which is a serious concern with Hi-C peak-calling. In simulation studies, HierCM showed equal or superior peak-calling accuracy compared to HiC-DC and Fit-Hi-C. HierCM also demonstrated better downsampling consistency and correlation with CTCF occupancy compared to HiC-DC on several datasets.