Single-cell RNA sequencing data analysis based on non-uniform ε-neighborhood network

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables the study of cellular heterogeneity and development at the individual cell level.
• Key computational challenges in scRNA-seq analysis include data visualization, cell clustering, and trajectory inference.
• Existing algorithms often lack a systematic or consistent approach to address these core analytical targets simultaneously.

Purpose of the Study:

• To develop an efficient and consistent computational framework for scRNA-seq data analysis.
• To integrate data visualization, cell clustering, and trajectory inference within a unified approach.
• To improve the accuracy and performance of scRNA-seq data analysis compared to existing methods.

Main Methods:

• A novel non-uniform ε-neighborhood (NEN) network is proposed to represent the data manifold in gene space.
• The NEN method combines advantages of k-nearest neighbors (KNN) and ε-neighborhood (EN) approaches.
• Network layout, community detection, and shortest path analysis are utilized for visualization, clustering, and trajectory inference.

Main Results:

• The NEN method accurately visualizes the global topological structure of datasets, outperforming t-SNE and UMAP.
• The framework demonstrates superior performance in cell clustering compared to existing approaches.
• The NEN method achieves more accurate pseudotime ordering of cells, enhancing trajectory inference.

Conclusions:

• The proposed NEN network framework offers a consistent and efficient solution for scRNA-seq data analysis.
• This integrated approach improves visualization, clustering, and trajectory inference capabilities.
• The ccnet Python package provides accessible implementation of this advanced analysis method.