Secuer

Secuer performs scalable and efficient spectral clustering of single-cell RNA sequencing (scRNA-seq) data using an anchor-based bipartite graph representation.

Key Features:

• Scalable and Efficient speCtral clUstERing algorithm: Implements a spectral clustering algorithm explicitly named "Scalable and Efficient speCtral clUstERing" for scRNA-seq data.
• Anchor-based bipartite graph representation: Uses an anchor-based bipartite graph representation algorithm to represent cells and anchors for clustering.
• Runtime and memory reduction: Reduces runtime and memory usage by over one order of magnitude for datasets containing more than 1 million cells.
• Benchmark accuracy: Maintains accuracy that is better or comparable to existing methods on small and moderate benchmark datasets.
• Consensus extension (Secuer-consensus): Serves as the foundation for Secuer-consensus, a consensus clustering extension that enhances scalability and runtime efficiency while preserving accuracy.
• Multi-scale applicability: Applicable to both small-scale and ultra-large single-cell clustering tasks.

Scientific Applications:

• scRNA-seq clustering: Clustering of single-cell RNA sequencing datasets, including ultra-large datasets exceeding one million cells.
• Consensus clustering workflows: Scaling and improving runtime efficiency of consensus clustering via Secuer-consensus.
• Transcriptomics studies: Identification of cell clusters and population structure in transcriptomics analyses.

Methodology:

Secuer constructs an anchor-based bipartite graph and applies a scalable spectral clustering algorithm, with an optional Secuer-consensus extension for consensus clustering.