scCAT: Single-cell Combined graph Attentional clustering for scRNA-seq analysis

Yufei Gao, Wenbo Zhang, Yameng Zhang, Yujie Shi, Lei Shi, Hailing Wang, Guozhen Cheng

Abstract: Highlights•This study focuses on capturing the complex nonlinear relationships in single-cell data while preserving key data variance to address challenges posed by high-dimensionality.•scCAT enhances learned representations by integrating gene expression and cellular topological relationships, which in turn improves the model’s ability to delineate cell subpopulations.•The results demonstrate superior cell subpopulation delineation performance of scCAT across eight scRNA-seq datasets, effectively mitigating the impact of inherent data defects.