SC-JNMF: single-cell clustering integrating multiple quantification methods based on joint non-negative matrix...

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) offers high-resolution biological insights.
• Accurate cell clustering is crucial for identifying cell types and diversity in scRNA-seq data.
• Current clustering methods are sensitive to variations introduced by different gene expression quantification pipelines.

Purpose of the Study:

• To develop a robust and accurate cell clustering method for single-cell expression analysis.
• To address the variability in clustering results caused by different quantification methods.
• To leverage information from multiple quantification methods for improved downstream analyses.

Main Methods:

• Proposed a novel clustering approach based on joint non-negative matrix factorization (joint-NMF).
• Joint-NMF integrates multiple gene expression profiles derived from the same RNA-seq data using different quantification methods.
• A shared factor matrix constraint is applied across multiple NMF instances to extract common biological signals.

Main Results:

• The joint-NMF method yields more robust and accurate cell clustering compared to conventional methods using single quantification profiles.
• Demonstrated improved identification of cell types, subpopulations, and biological diversity.
• Showcased the utility of extracted features from joint-NMF for discovering novel marker genes.

Conclusions:

• Integrating data from multiple quantification methods using joint-NMF enhances the reliability of single-cell RNA sequencing analysis.
• This approach mitigates biases associated with individual quantification techniques.
• The method provides a powerful tool for precise cell type classification and marker gene discovery in complex biological systems.