sc-REnF: An entropy guided robust feature selection for single-cell RNA-seq data.

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) data analysis relies on accurate cell population annotation.
• Technical noise in scRNA-seq data necessitates robust gene selection for reliable downstream analysis, particularly for clustering.
• Existing gene selection methods may struggle with the inherent noise and high dimensionality of scRNA-seq data.

Purpose of the Study:

• To develop a robust gene selection method for single-cell clustering.
• To leverage Renyi and Tsallis entropies for improved feature selection in noisy scRNA-seq data.
• To enhance the accuracy and performance of single-cell clustering.

Main Methods:

• Introduction of sc-REnF (robust entropy based feature selection method).
• Utilizing Renyi and Tsallis entropies for gene selection, with a tunable parameter (q).
• Evaluation of sc-REnF against competing methods on scRNA-seq datasets.

Main Results:

• sc-REnF significantly improved clustering results compared to other methods.
• The method effectively captures relevancy and redundancy in noisy data due to its robust objective function.
• Selected genes accurately identified unknown cell populations, demonstrating high predictive power.
• sc-REnF achieved good clustering performance on small sample, large feature scRNA-seq data.

Conclusions:

• sc-REnF offers a robust and effective approach for gene selection in single-cell clustering.
• The method enhances the reliability of cell annotation by improving clustering performance.
• sc-REnF is particularly beneficial for analyzing noisy and high-dimensional scRNA-seq data.