scDMSC: Deep Multi-View Subspace Clustering for Single-Cell Multi-Omics Data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell multi-omics sequencing offers comprehensive cellular insights but faces challenges with high dimensionality, sparsity, and heterogeneity.
• Clustering analysis of multi-omics data is crucial for understanding cellular heterogeneity and biological mechanisms.

Purpose of the Study:

• To develop an unsupervised clustering algorithm for effective single-cell multi-omics data integration.
• To address the challenges of high dimensionality, sparsity, and heterogeneity in multi-omics datasets.
• To uncover shared latent features and correlations among different omics data.

Main Methods:

• Proposes scDMSC, an unsupervised clustering algorithm based on deep multi-view subspace learning.
• Utilizes weighted reconstruction to manage omics data heterogeneity.
• Employs deep subspace learning to identify shared latent features and inter-omics correlations.

Main Results:

• scDMSC demonstrates superior performance in precision and scalability compared to existing methods on real and simulated datasets.
• The algorithm effectively integrates diverse omics data, revealing cellular heterogeneity.
• Downstream analyses confirm the model's ability to uncover biological mechanisms through differential expression and modality interpretability.

Conclusions:

• scDMSC provides a robust and scalable solution for single-cell multi-omics data clustering and integration.
• The method enhances the understanding of cellular complexity and biological mechanisms.
• This approach advances the field of single-cell multi-omics data analysis.