Cell Type Differentiation Using Network Clustering Algorithms.

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 data analysis.
• In-silico algorithms are essential for understanding cellular heterogeneity and disease mechanisms.

Purpose of the Study:

• To compare the performance of various algorithms for cell type identification in scRNA-seq data.
• To analyze and optimize preprocessing pipelines for scRNA-seq analysis.
• To evaluate clustering methods on cell-cell networks derived from gene expression data.

Main Methods:

• Comparative analysis of clustering algorithms: Seurat, Leiden, WGCNA, Infomap, Stochastic Block Models (SBM), and single-cell Graph Neural Networks (scGNN).
• Analysis of preprocessing pipelines for scRNA-seq data.
• Application of clustering algorithms to cell-cell networks derived from PBMC and ROSMAP datasets.

Main Results:

• Clustering by WGCNA showed limited correspondence with known cell types.
• Multiresolution Infomap, Leiden, and SBM algorithms demonstrated closer alignment with cell types.
• Infomap emerged as a highly effective approach for cell type identification.

Conclusions:

• Infomap and Leiden algorithms provide robust cell type annotation in scRNA-seq data.
• These methods are valuable for characterizing cellular landscapes in neurodegeneration and immunology.
• Optimized preprocessing and clustering are key to unlocking scRNA-seq's potential.