Branch-specific gene discovery in cell differentiation using multi-omics graph attention.

Area of Science:

• Genomics
• Computational Biology
• Cell Biology

Background:

• Cell differentiation involves complex gene regulation.
• Integrating multi-omics single-cell data is crucial for understanding these dynamics.
• Existing methods may not fully capture branch-specific gene importance.

Purpose of the Study:

• To develop a framework for identifying branch-specific key genes during cell differentiation using multi-omics single-cell data.
• To integrate single-cell RNA sequencing (scRNA-seq) and single-cell Assay for Transposase-Accessible Chromatin sequencing (scATAC-seq) data.
• To provide a tool for dissecting gene regulatory networks and cell fate decisions.

Main Methods:

• Proposed BranchKGN, a heterogeneous graph transformer-based framework.
• Integrated scRNA-seq and scATAC-seq data into a unified gene representation.
• Inferred differentiation trajectories using Slingshot and constructed a heterogeneous graph.
• Employed attention-based graph learning to assign gene importance scores.

Main Results:

• Identified genes consistently informative across cell differentiation branches.
• Successfully reconstructed gene regulatory networks and differentiation trajectories.
• Validated the framework on three independent datasets, demonstrating accurate identification of key regulators.
• Showcased the ability to capture key regulators of cell fate bifurcation.

Conclusions:

• BranchKGN effectively dissects gene regulation dynamics during cellular transitions.
• The framework provides a valuable tool for multi-omics single-cell analysis.
• Identified gene sets accurately reconstruct differentiation trajectories and capture key regulatory events.