Ultrafast and interpretable single-cell 3D genome analysis with Fast-Higashi

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Single-cell Hi-C (scHi-C) technology enables the study of 3D genome organization within individual cells.
• Existing scHi-C analysis methods face challenges due to data sparsity, quality issues, complex 3D genome patterns, and limited computational scalability and interpretability.
• These limitations hinder the comprehensive analysis of large-scale scHi-C datasets and the accurate identification of distinct cellular states.

Purpose of the Study:

• To develop an ultrafast and interpretable computational method for analyzing sparse scHi-C data.
• To enable the joint identification of cell identities and chromatin meta-interactions from scHi-C data.
• To overcome the limitations of existing methods in terms of speed, interpretability, and scalability for large-scale single-cell 3D genomics.

Main Methods:

• Introduction of Fast-Higashi, a novel method employing tensor decomposition and partial random walk with restart.
• Application of Fast-Higashi to analyze sparse scHi-C data for joint cell identity and chromatin interaction identification.
• Extensive evaluations comparing Fast-Higashi against existing computational tools for scHi-C data analysis.

Main Results:

• Fast-Higashi demonstrates superior performance compared to existing methods in analyzing scHi-C data.
• The method significantly improves the delineation of rare cell types and continuous developmental trajectories.
• Fast-Higashi successfully identifies 3D genome features defining distinct cell types and elucidates cell-type-specific genome structure-function relationships.

Conclusions:

• Fast-Higashi offers a highly efficient and interpretable solution for scHi-C data analysis.
• The method enhances the discovery of cellular heterogeneity and dynamic biological processes through 3D genome structure analysis.
• Fast-Higashi is broadly applicable across diverse biological contexts and can integrate with other single-cell omics data.