Bayesian inference of chromatin structure ensembles from population-averaged contact data

Area of Science:

• Genomics
• Molecular Biology
• Computational Biology

Background:

• Chromatin's spatial organization influences nuclear processes like transcription.
• Chromosome conformation capture (3C) techniques map genome-wide contacts.
• Population-averaged 3C data obscure single-cell 3D genome organization and variability.

Purpose of the Study:

• To develop a computational method for inferring ensembles of chromatin structures from 3C data.
• To objectively determine the optimal number of structural states within a cell population.
• To analyze cell-to-cell variability in 3D genome organization.

Main Methods:

• A fully Bayesian statistical framework for chromatin structure modeling.
• Inference of multi-state chromatin models from chromosome conformation capture carbon copy (5C) data.
• Validation using simulated data and comparison with independent experimental data.

Main Results:

• The proposed method successfully infers ensembles of chromatin structures.
• Multi-state models were computed from 5C data, reflecting population heterogeneity.
• Inferred ensembles accurately represent the underlying cell population.

Conclusions:

• The Bayesian approach captures cell-to-cell variability in 3D genome organization.
• This method provides insights into chromatin structure dynamics and topologically associating domains.
• The approach enhances understanding of transcription regulation through spatial genome organization.