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

## TL;DR

This paper introduces BranchKGN, a new method that uses multi-omics data to identify key genes involved in cell differentiation and reconstruct gene regulatory networks.

## Contribution

The novel contribution is a graph attention-based framework for identifying branch-specific key genes in cell differentiation using scRNA-seq and scATAC-seq data.

## Key findings

- BranchKGN successfully identifies key genes at bifurcation points in cell differentiation trajectories.
- The identified genes support accurate reconstruction of differentiation trajectories across multiple datasets.

## Abstract

Understanding gene regulation during cell differentiation requires effective integration of multi-omics single-cell data. In this study, we propose BranchKGN, a heterogeneous graph transformer-based framework for identifying branch-specific key genes along cell differentiation trajectories. By integrating scRNA-seq and scATAC-seq data into a unified gene representation, we infer differentiation trajectories using Slingshot and construct a heterogeneous graph capturing gene–cell relationships. Through attention-based graph learning, BranchKGN assigns gene importance scores within each cell, enabling the identification of genes consistently informative across branch point cells and their descendant lineages. These genes are then used to reconstruct gene regulatory networks and differentiation trajectories. Validation on three independent datasets demonstrates that the identified gene sets not only capture key regulators of cell fate bifurcation but also support accurate reconstruction of differentiation trajectories. Our results highlight the effectiveness of BranchKGN in dissecting gene regulation dynamics during cellular transitions and provide a valuable tool for multi-omics single-cell analysis.

Cell fate is an important biology process during biological development, tissue regeneration, and disease progression. However, the modes of cell differentiation and the patterns of gene expression changes during this process are still not so clear. To investigate these problems, we introduce BranchKGN, a framework based on graph attention mechanisms designed to identify branch-specific key genes along cell differentiation trajectories. By integrating scRNA-seq and scATAC-seq data and inferring differentiation trajectories using Slingshot, BranchKGN employs multi-head attention learning to score the importance of genes within each cell. And then branch-specific key genes at bifurcation points are identified. These genes are used to reconstruct gene regulatory networks and differentiation trajectories. We test BranchKGN on three independent datasets, can identify specific key gene sets at cell fate bifurcations. These key genes can accurately reconstruct the cell differentiation trajectories in turn. Our results highlight that BranchKGN is a powerful tool to dissect gene regulatory dynamics during cell transitions and analyze the multi-omics single-cell data.