SC3s

SC3s performs consensus clustering of large-scale single-cell transcriptomic and single-cell RNA sequencing datasets using a streaming k-means approach to identify cell types and states.

Key Features:

• Consensus clustering: Implements consensus clustering for unsupervised identification of cell populations.
• Scalability: Processes datasets containing up to 2 million cells.
• Efficiency: Employs a streaming k-means clustering algorithm optimized for time and memory efficiency.
• Streaming processing: Uses a streaming algorithm that enables continuous processing of large datasets.
• Unsupervised clustering: Focused on unsupervised clustering to identify cell types and states without prior labels.

Scientific Applications:

• Single-cell transcriptomics: Enables analysis of large single-cell transcriptomic datasets to reveal cellular heterogeneity and complex biological processes.
• Biomedical research: Supports applications including disease modeling, developmental biology, and cancer studies through scalable clustering of single-cell data.

Methodology:

SC3s uses a k-means clustering algorithm implemented in a streaming fashion and applies consensus clustering; the approach is optimized for time and memory and scales favorably with the number of cells.