Clustering single-cell multimodal omics data with jrSiCKLSNMF.

Area of Science:

• Single-cell biology
• Multi-omics data integration
• Computational biology

Background:

• Multimodal single-cell omics methods generate rich data from individual cells.
• Integrating diverse omics data offers deeper insights into cellular functions.
• Single-cell omics data present challenges like high dimensionality, sparsity, and noise.

Purpose of the Study:

• To introduce a novel multimodal data analysis method for single-cell omics.
• To address the challenges of modeling complex single-cell omics data.
• To improve the accuracy of cell type and function identification.

Main Methods:

• We propose joint graph-regularized Single-Cell Kullback-Leibler Sparse Non-negative Matrix Factorization (jrSiCKLSNMF).
• This method extracts shared latent factors across multiple omics modalities from the same cells.
• The approach is designed to handle high dimensionality and sparsity inherent in single-cell data.

Main Results:

• jrSiCKLSNMF demonstrated superior clustering performance compared to existing methods on simulated datasets.
• The algorithm was applied to a real-world cell line dataset.
• Clustering results on the real data were found to be scientifically accurate.

Conclusions:

• jrSiCKLSNMF offers a powerful new tool for multimodal single-cell data analysis.
• The method effectively integrates information across different omics modalities.
• This approach enhances the understanding of cellular heterogeneity and function.