Title: “Bayesian inference of multi-state models for 3D genome organization from chromosome conformation capture data”
Abstract: In recent years, genome-wide chromosome conformation capture (3C) techniques such as 5C and Hi-C have given great insight in the detailed three-dimensional organization of genomes. Usually, populations of cells are analyzed and the resulting data thus contains information about the three-dimensional structure of millions of molecules, which complicates modeling these structures as compared to modeling from single-cell data.
We present a Bayesian approach to genome structure determination which infers multi-state structural models from 5C and Hi-C data, thereby demixing the population-averaged data. Leveraging many advantages of a fully Bayesian analysis, we obtain not only a statistically well-defined measure of uncertainty for our multi-state models, but also estimate important modeling parameters such as the optimal number of states from the data. While this presentation will be focused on methods, we also discuss several biologically relevant questions which our approach might help to answer.