Extracting physical characteristics of higher-order chromatin structures from 3D image data

Area of Science:

• Genomics
• Computational Biology
• Molecular Imaging

Background:

• Higher-order chromatin structures influence gene regulation and cell identity.
• High-throughput sequencing (HTS) methods like Hi-C analyze chromatin structure at the population level.
• High-resolution 3D imaging techniques offer single-cell chromatin structure insights, but computational analysis remains challenging.

Purpose of the Study:

• To develop a computational method for analyzing high-resolution 3D chromatin images.
• To enable explainable and interpretable analysis of chromatin structures and their functional characteristics.
• To identify physical features defining open or closed chromatin domains in single cells.

Main Methods:

• Developed Extracting Physical-Characteristics from Images of Chromatin Structures (EPICS), a machine learning-based computational method.
• Applied EPICS to 3D image data generated by 3D electron microscopy with in situ hybridization (3D-EMSIH) and 3D structured illumination microscopy (3D-SIM).
• Utilized EPICS to generate 3D chromatin representations, identify domains, and determine their open/closed status.

Main Results:

• EPICS successfully processed 3D chromatin image data, providing direct 3D structural representations.
• The method identified major chromatin domains and classified them as open or closed.
• EPICS highlighted key features contributing to domain status, demonstrating model explainability and interpretability.

Conclusions:

• EPICS provides a robust computational framework for analyzing single-cell 3D chromatin structures from high-resolution imaging.
• The tool enhances understanding of chromatin organization and its functional implications.
• EPICS is applicable to diverse spatial genomics studies utilizing 3D molecular imaging data.