scSO

scSO performs unsupervised clustering of single-cell RNA sequencing (scRNA-seq) data using sparse optimization and low-rank matrix factorization to identify cell-type structure.

Key Features:

• Sparse Optimization and Low-Rank Matrix Factorization: The core methodology leverages sparse optimization combined with low-rank matrix factorization to decompose scRNA-seq matrices into interpretable components for clustering.
• Unsupervised Clustering: scSO operates without prior cell-type labels, enabling exploratory analysis of heterogeneous scRNA-seq datasets.
• Accuracy and Validation: Validated on multiple benchmark datasets, scSO's predicted cluster numbers closely align with known reference cell types.

Scientific Applications:

• Cell Type Identification: By clustering scRNA-seq data, scSO distinguishes cell types within heterogeneous samples.
• Biomarker Discovery: Precise cell classification facilitates identification of biomarkers associated with specific cell types or states in oncology, immunology, and developmental biology.
• Comparative Analysis Across Biological Systems: Applicable to human tissues and model organisms for comparative and cross-species analyses.

Methodology:

scSO applies sparse optimization to identify key features and then employs low-rank matrix factorization for dimensionality reduction while preserving essential structural information to extract meaningful clusters.