FEATS: feature selection-based clustering of single-cell RNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Next-generation sequencing enables single-cell RNA sequencing (scRNA-seq) for high-resolution transcriptomic studies.
• The rapid generation of large scRNA-seq datasets necessitates advanced analytical tools for improved accuracy and efficiency.

Purpose of the Study:

• To introduce FEATS, a Python software package designed for clustering scRNA-seq data.
• To enhance the accuracy and efficiency of scRNA-seq data analysis through a novel feature selection approach.

Main Methods:

• FEATS employs a univariate feature selection-based approach for clustering.
• This method identifies top informative features, mimicking manual cell type determination using marker genes.
• The package integrates clustering, outlier detection, and data integration capabilities.

Main Results:

• FEATS demonstrates superior performance compared to existing tools on various scRNA-seq datasets.
• It achieves a 22% improvement in clustering accuracy, measured by the adjusted Rand index.
• FEATS also excels in accurately estimating the number of clusters and offers excellent computational performance.

Conclusions:

• FEATS is a comprehensive and efficient tool for scRNA-seq data clustering.
• Its feature selection approach improves cell type identification and data integration.
• The software package addresses key challenges in analyzing scRNA-seq data.