scLGGCL: Label-Guided Graph Contrastive Learning for Single-Cell Fusion Clustering

Single-cell RNA sequencing (scRNA-seq) provides transcriptomic profiles at cellular resolution, enabling the study of tissue heterogeneity. A critical step in scRNA-seq analysis is cell clustering, which identifies distinct subpopulations for downstream biological interpretation. However, most existing clustering algorithms fail to simultaneously leverage cellular attributes and intercellular structural relationships. Additionally, graph-based methods that employ contrastive learning typically neglect cell-level semantic similarity. We present scLGGCL, a label-guided graph contrastive learning approach to tackle the above limitations. The framework integrates three modules: dual-reconstruction to fuse attribute-structure information, contrastive learning under label guidance to extract semantic similarities, and deep embedding clustering to enable iterative optimization. Comprehensive evaluations on single and cross-dataset benchmarks show that scLGGCL achieves superior clustering performance. Code is available at https://github.com/CDMBlab/scLGGCL .