NonLoss: a novel analytical method for differential biological module identification from single-cell transcriptome.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Identifying disease-related biological modules is crucial for understanding disease mechanisms.
• Single-cell RNA sequencing (scRNA-seq) offers high-resolution transcriptome data but is susceptible to individual gene expression noise, complicating module identification.

Purpose of the Study:

• To develop a robust computational method for analyzing scRNA-seq data.
• To accurately identify biological function modules and subtle gene expression changes at single-cell resolution.

Main Methods:

• Developed NonLoss, a computational method and desktop application.
• Utilized Shannon's entropy and Spearman rank correlation for variation and consistency analysis.
• Applied dimensionality reduction and reliability analysis on all genes within a function module.

Main Results:

• NonLoss robustly analyzes scRNA-seq data, distinguishing real biological differences between cell populations.
• The tool effectively handles expression data from all genes in a module, reducing random gene influence.
• Demonstrated biologically significant insights in three distinct applications, identifying key genes with subtle expression changes.

Conclusions:

• NonLoss provides a user-friendly graphical interface for scRNA-seq data analysis.
• The tool successfully identifies biologically relevant expression changes at the single-cell level.