Sc-GPE: A Graph Partitioning-Based Cluster Ensemble Method for Single-Cell.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) analysis relies on clustering to identify cell types and understand differentiation.
• Existing clustering methods yield variable results, complicating biological interpretation.
• Cluster ensemble strategies offer a solution to consolidate diverse clustering outcomes.

Purpose of the Study:

• To develop a novel cluster ensemble method, Sc-GPE, for improved scRNA-seq data analysis.
• To address challenges in integrating results from different graph partitioning-based clustering algorithms.
• To enhance the accuracy and biological relevance of cell type identification from scRNA-seq data.

Main Methods:

• Sc-GPE combines five single-cell graph partitioning methods: SNN-cliq, PhenoGraph, SC3, SSNN-Louvain, and MPGS-Louvain.
• A consensus matrix is built by calculating cell-pair clustering probabilities, overcoming label assignment discrepancies.
• A weighted consensus matrix incorporates method importance scores, followed by hierarchical clustering.

Main Results:

• Sc-GPE demonstrated superior average performance compared to individual methods and SAME-clustering across 12 scRNA-seq datasets.
• The method achieved the highest Normalized Mutual Information (NMI) and Adjusted Rand Index (ARI) on five datasets.
• Sc-GPE effectively integrates clustering solutions, improving cell type identification accuracy.

Conclusions:

• Sc-GPE provides a robust and effective cluster ensemble approach for scRNA-seq data.
• The method enhances the reliability of cell type identification and biological interpretation.
• Sc-GPE represents a significant advancement in analyzing complex single-cell transcriptomic data.