scMCs: a framework for single-cell multi-omics data integration and multiple clusterings.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell multi-omics data integration reveals cellular regulatory mechanisms.
• Current methods struggle with data noise, sparsity, heterogeneity, and omics individuality.
• Existing clustering methods lack the ability to discover alternative cell groupings.

Purpose of the Study:

• To introduce a novel single-cell data fusion approach for multiple clustering (scMCs).
• To jointly model single-cell transcriptomics and epigenetic data.
• To explore and uncover alternative cell clusterings beyond traditional cell-type identification.

Main Methods:

• scMCs mines omics-specific and cross-omics representations.
• Data is fused into a co-embedding representation for heterogeneity dissection and imputation.
• Alternative clusterings are discovered by projecting into salient subspaces, reducing redundancy, and optimizing cluster centers.

Main Results:

• scMCs effectively identifies subcellular types and imputes data dropout events.
• The approach uncovers diverse cell characteristics through multiple, meaningful clusterings.
• Experimental results demonstrate the efficacy of scMCs in analyzing complex single-cell multi-omics data.

Conclusions:

• scMCs provides a powerful framework for single-cell multi-omics data analysis.
• The ability to generate multiple clusterings offers deeper insights into cellular heterogeneity.
• This method enhances our understanding of complex genetic information within cells.