ClusterMatch

ClusterMatch aligns single-cell RNA sequencing data at the multi-scale cluster level using a stable matching optimization model to improve data integration and cell type annotation.

Key Features:

• Canonical Correlation Analysis: ClusterMatch leverages canonical correlation analysis to capture mutual correspondence between datasets in a latent space.
• Multi-scale Louvain Clustering: ClusterMatch employs multi-scale Louvain clustering algorithms to identify clusters with optimized resolutions across varying scales.
• Stable Matching Framework: ClusterMatch utilizes a stable matching framework to align scRNA-seq data while maintaining interpretability.
• Marker Gene Set Integration: ClusterMatch incorporates overlapped marker gene sets to provide interpretability in the alignment results.
• Global and Local Information Utilization: ClusterMatch utilizes both global and local information of scRNA-seq data for alignment.

Scientific Applications:

• Data Integration: ClusterMatch integrates multiple scRNA-seq datasets through cluster-level alignment in the latent space.
• Cell Type Annotation: ClusterMatch annotates cell types by aligning clusters and leveraging marker gene sets for interpretability.
• Cross-species Alignment: ClusterMatch aligns scRNA-seq data across different species at the cluster level.
• Cross-timepoint Alignment: ClusterMatch aligns scRNA-seq data across different timepoints to characterize cellular dynamics.

Methodology:

ClusterMatch applies canonical correlation analysis and multi-scale Louvain clustering to identify clusters with optimized resolutions, then uses a stable matching framework to align scRNA-seq data in the latent space with interpretability maintained through overlapped marker gene sets.