Network clustering algorithms and preprocessing pipelines for robust cell type identification in single-cell RNA...

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) offers high-resolution transcriptomic data.
• Accurate cell type annotation is a critical challenge in scRNA-seq analysis.
• Identifying cell types is essential for understanding cellular heterogeneity and disease mechanisms.

Purpose of the Study:

• To compare the performance of various computational methods for cell type identification in scRNA-seq data.
• To analyze preprocessing pipelines for optimizing noise and batch effect mitigation.
• To identify robust algorithms for accurate cell type annotation.

Main Methods:

• Comparative analysis of established (Seurat, Leiden, WGCNA) and network-based (Infomap, SBM, scGNN) algorithms.
• Evaluation of preprocessing pipeline components for noise and batch effect reduction.
• Clustering of cell-cell networks derived from gene expression data across three independent datasets (PBMC, ROSMAP, MOp).

Main Results:

• Multiresolution Infomap and Leiden clustering algorithms demonstrated strong alignment in identifying cell types.
• Infomap emerged as a highly effective method for robust cell type identification.
• Preprocessing pipeline optimization was analyzed for improved data quality.

Conclusions:

• Network-based clustering, particularly Infomap, provides a robust approach for cell type annotation in scRNA-seq.
• Infomap offers significant advantages for characterizing cellular landscapes in immunology and neurodegeneration research.
• Optimized preprocessing and network analysis are key to unlocking the full potential of scRNA-seq data.