Intrinsic entropy model for feature selection of scRNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) reveals cellular heterogeneity but suffers from noise and dropout events.
• Accurate downstream analysis, particularly clustering, relies heavily on effective feature gene selection.

Purpose of the Study:

• To develop a novel feature selection method for scRNA-seq data using an intrinsic entropy (IE) model.
• To identify informative genes that improve the accuracy of clustering analysis by distinguishing biological variation from technical noise.

Main Methods:

• Derivation of an entropy decomposition formula to separate total entropy (TE) into intrinsic entropy (IE) and extrinsic entropy (EE).
• Implementation of the IE model to extract gene-specific IE, representing regulatory fluctuations.
• Computational validation using simulated and real scRNA-seq datasets, comparing performance against existing methods.

Main Results:

• The IE model successfully identifies informative genes, demonstrating high performance in clustering and classification tasks.
• High-IE genes reflect regulatory fluctuations crucial for cell type and state identification.
• The IE model shows broad applicability, robustness across different analytical methods, and sensitivity for detecting novel cell types.

Conclusions:

• Intrinsic entropy (IE) represents valuable biological information, not noise, in scRNA-seq data.
• The proposed IE model enhances the accuracy and reliability of scRNA-seq data analysis, particularly for clustering.
• This method offers a robust approach for leveraging gene expression variability to understand cellular states.