MOJITOO: a fast and universal method for integration of multimodal single-cell data.

Area of Science:

• Single-cell biology
• Computational biology
• Genomics

Background:

• Multi-modal single-cell sequencing allows simultaneous inspection of transcriptomes, epigenomes, and proteomes.
• Existing computational methods for multimodal single-cell data integration are often computationally expensive, require parameter tuning, or are modality-specific.

Purpose of the Study:

• To present a novel, efficient, and parameter-free computational method for integrating multimodal single-cell data.
• To enable robust detection and interpretation of shared cellular representations across different molecular modalities.

Main Methods:

• Developed Multi-mOdal Joint IntegraTion of cOmpOnents (MOJITOO), a method utilizing canonical correlation analysis.
• MOJITOO identifies a shared latent space from multimodal single-cell data without requiring parameter delineation.
• Canonical components derived from MOJITOO facilitate the association of modality-specific features with the latent space.

Main Results:

• MOJITOO demonstrates superior performance in terms of computational requirements compared to existing methods.
• The method effectively preserves the original latent spaces of the single-cell data.
• MOJITOO shows improved clustering performance on bi- and tri-modal single-cell datasets.

Conclusions:

• MOJITOO offers a computationally efficient and parameter-free solution for multimodal single-cell data integration.
• The method facilitates the interpretation of complex single-cell data by linking molecular features to a shared representation.
• MOJITOO represents a significant advancement in the analysis of multi-omic single-cell datasets.