scLM: Automatic Detection of Consensus Gene Clusters Across Multiple Single-cell Datasets.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Gene co-expression analysis is crucial for understanding cell identity and function in gene expression profiling.
• Existing methods for single-cell RNA sequencing (scRNA-seq) data struggle with accurate co-expressed gene identification.
• Current approaches often yield results that limit biological interpretation of co-expressed gene networks.

Purpose of the Study:

• To introduce single-cell Latent-variable Model (scLM), a novel algorithm for gene co-clustering in scRNA-seq data.
• To address the limitations of current methods in accurately detecting biologically relevant gene clusters.
• To enable consensus clustering across multiple scRNA-seq datasets for comparative analysis.

Main Methods:

• Developed scLM, a gene co-clustering algorithm specifically designed for single-cell data.
• scLM processes raw count data, preserving biological variation and mitigating batch effects.
• Implemented consensus clustering to integrate multiple scRNA-seq datasets.

Main Results:

• scLM accurately identifies gene clusters with significant biological context.
• The algorithm demonstrates superior performance compared to existing methods in both simulated and experimental data.
• scLM successfully identified novel functional gene modules and refined cell states in experimental datasets.

Conclusions:

• scLM offers a robust and accurate approach for gene co-expression analysis in scRNA-seq data.
• The method facilitates mechanism discovery and enhances the understanding of complex biological systems, including cancers.
• A user-friendly R package for scLM is publicly available, promoting wider adoption and application.