MOJITOO

MOJITOO applies canonical correlation analysis (CCA) to integrate multimodal single-cell data by identifying shared latent components across transcriptomic, epigenomic, and proteomic modalities.

Key Features:

• Canonical Correlation Analysis (CCA): Uses CCA to detect shared representations across different cellular modalities.
• Latent Component Identification: Identifies latent components that encapsulate commonalities among transcriptomes, epigenomes, and proteomes.
• Preservation of Original Latent Spaces: Maintains the integrity of original latent spaces during integration to preserve biological variability.
• Parameter Requirements: Performs integration without requiring predefined parameters.
• Computational Efficiency: Demonstrates improved computational efficiency and scalability relative to some existing methods.
• Interpretability: Produces canonical components that can be used to associate specific molecular features with the integrated latent space.
• Multimodal Applicability: Applicable to bi-modal and tri-modal single-cell datasets.
• Clustering Performance: Shows improved clustering performance in comparative analyses.

Scientific Applications:

• Multimodal single-cell integration: Integration and joint analysis of transcriptomic, epigenomic, and proteomic single-cell datasets.
• Systems Biology: Studying interactions and shared signals across molecular layers within complex biological systems.
• Developmental Biology: Investigating cell-state transitions and lineage relationships using integrated modalities.
• Disease Modeling: Elucidating disease-associated molecular signatures across multiple omics layers at single-cell resolution.

Methodology:

Applies canonical correlation analysis (CCA) to derive canonical components that represent shared latent structures across modalities and identifies latent components while preserving original latent spaces.