MOH: A Novel Multilayer Multi-Omics Heterogeneous Graph for Single-Cell Clustering.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Cell clustering is vital for identifying cell populations in single-cell multi-omics research.
• Integrating more than two omics types for clustering is challenging due to data diversity.
• Existing methods using heterogeneous graphs have limitations in capturing cell interactions and scalability.

Purpose of the Study:

• To develop a novel single-cell clustering algorithm, MOH, capable of integrating multiple omics data types.
• To address the limitations of traditional methods in handling data heterogeneity and scalability.
• To improve the accuracy and comprehensiveness of cell clustering in multi-omics studies.

Main Methods:

• Introduced MOH, a single-cell clustering algorithm based on a multilayer multi-omics heterogeneous graph.
• Integrated three omics types: single-cell RNA sequencing (scRNA-seq), single-cell ATAC sequencing (scATAC-seq), and spatial transcriptomics.
• Constructed a multilayer heterogeneous graph to extract and enhance representations from all omics layers, capturing intra-layer and inter-layer relationships.

Main Results:

• MOH demonstrated superior performance compared to six state-of-the-art methods on unsupervised clustering metrics.
• The algorithm achieved consistent improvements across all evaluation criteria, indicating precise and comprehensive analysis.
• Downstream analyses revealed novel biological insights into cancer complications, drug repurposing, and signaling pathways.

Conclusions:

• MOH provides an effective approach for single-cell clustering by integrating multiple omics data through a multilayer heterogeneous graph.
• The method enhances the ability to identify distinct cellular populations and uncover complex biological relationships.
• The findings suggest significant potential for MOH in advancing multi-omics research and biological discovery.