scME: a dual-modality factor model for single-cell multiomics embedding.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Single-cell multiomics technologies enable the characterization of diverse molecular features within cells.
• Integrating these multi-modal data is crucial for dissecting cellular heterogeneity.
• Current methods often prioritize shared information, neglecting modality-specific complementary data.

Purpose of the Study:

• To develop a method that disentangles and integrates both shared and complementary information across multiple molecular modalities.
• To improve the joint representation of single-cell multi-omics data.
• To enhance downstream analyses such as single-cell clustering and cell-type classification.

Main Methods:

• Development of a dual-modality factor model named scME.
• Utilizing deep factor modeling for data integration.
• Evaluating scME's performance against existing single-cell multiomics integration algorithms.

Main Results:

• scME generates a superior joint representation of multiple modalities compared to other methods.
• The model effectively elucidates nuanced differences among cells.
• The joint representation from scME significantly improves single-cell clustering and cell-type classification accuracy.

Conclusions:

• scME is an efficient method for integrating diverse molecular features from single-cell multiomics data.
• The approach facilitates a more comprehensive dissection of cell heterogeneity.
• scME provides a valuable tool for advancing single-cell multiomics research.