A cofunctional grouping-based approach for non-redundant feature gene selection in unannotated single-cell RNA-seq...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Feature gene selection is crucial for cell clustering in single-cell RNA sequencing (scRNA-seq) analysis.
• Existing methods often overlook gene redundancy and complementarity, impacting clustering accuracy.
• A comprehensive feature selection approach is needed to optimize scRNA-seq data analysis.

Purpose of the Study:

• To develop a novel computational method, GeneClust, for effective feature gene selection in scRNA-seq cell clustering.
• To address the limitations of existing methods by incorporating gene relevance, redundancy, and complementarity.
• To provide a versatile tool that enhances cell clustering and facilitates biological interpretation.

Main Methods:

• GeneClust groups genes based on expression profiles.
• It selects feature genes by maximizing relevance, minimizing redundancy, and preserving complementarity.
• The method is designed as a plug-in tool compatible with existing clustering algorithms.

Main Results:

• GeneClust significantly improves cell clustering performance in scRNA-seq data.
• Benchmark results confirm the method's effectiveness compared to existing approaches.
• GeneClust successfully groups co-functional genes, aiding in the investigation of gene interactions and biological pathways.

Conclusions:

• GeneClust offers a superior approach to feature gene selection for scRNA-seq cell clustering.
• The method not only enhances clustering accuracy but also provides valuable biological insights.
• GeneClust is a freely available computational tool with broad applicability in transcriptomic studies.