A method for chromatin domain partitioning based on hypergraph clustering.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Multi-scale chromatin domain models (A/B compartments, TADs) are widely used.
• Existing methods struggle to identify structures across multiple scales simultaneously.
• A unified approach for multi-scale chromatin domain partitioning is needed.

Purpose of the Study:

• To develop a novel method, TORNADOES, for comprehensive chromatin domain partitioning across multiple scales.
• To integrate Hi-C and ChIP-seq data for enhanced identification of genomic structures.
• To provide a deeper understanding of the hierarchical organization and functional differences within TADs.

Main Methods:

• Utilized a density clustering algorithm on Hi-C data at varying resolutions to identify TADs.
• Constructed a hypergraph integrating ChIP-seq features with multi-scale TAD data.
• Applied hypergraph Laplacian matrix analysis for partitioning chromatin domains (A/B, A1-B3).

Main Results:

• TORNADOES successfully identified chromatin domains with distinct features at multiple scales.
• Experiments confirmed the method's ability to reveal organizational patterns and functional differences within TADs.
• Comparative analyses showed TORNADOES outperforms other methods in classifying compartment types and numbers.

Conclusions:

• TORNADOES offers a robust framework for multi-scale chromatin domain partitioning.
• The method enhances understanding of genome hierarchical structure and functional genomics.
• TORNADOES provides a valuable tool for analyzing complex genomic organization across different cell types.