DEMOC: a deep embedded multi-omics learning approach for clustering single-cell CITE-seq data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables cell-type identification through clustering.
• Cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) captures both RNA and protein expression.
• Existing clustering algorithms struggle with multi-omics data integration.

Purpose of the Study:

• To develop a novel deep embedded multi-omics clustering (DEMOC) model for joint clustering of CITE-seq data.
• To effectively integrate transcriptomic and proteomic data for enhanced cell analysis.
• To improve cell-type identification and analysis of cellular heterogeneity.

Main Methods:

• Proposed a deep embedded multi-omics clustering with collaborative training (DEMOC) model.
• Designed the model to leverage consistent and complementary information from transcriptomic and proteomic data.
• Validated the model on CITE-seq and scRNA-seq datasets.

Main Results:

• DEMOC outperforms state-of-the-art single-omic clustering methods on CITE-seq data.
• DEMOC achieves superior and stable performance compared to existing multi-omics clustering methods.
• The model demonstrates effectiveness in rare cell-type identification, novel cell-subtype detection, and cellular heterogeneity analysis on scRNA-seq data.

Conclusions:

• The DEMOC model provides an effective approach for joint clustering of multi-omics single-cell data.
• DEMOC enhances cell-type identification and analysis of cellular heterogeneity.
• This method offers a powerful tool for analyzing complex single-cell datasets.