CMOT: Cross-Modality Optimal Transport for multimodal inference.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Multimodal single-cell sequencing provides deep biological insights.
• Integrating diverse single-cell data modalities is challenging due to missing data and cell correspondences.

Purpose of the Study:

• To develop a novel computational approach for seamless integration of multimodal single-cell data.
• To address the limitations of existing methods in handling missing modalities and cell mapping.

Main Methods:

• Developed Cross-Modality Optimal Transport (CMOT), a computational framework.
• CMOT aligns cells into a common latent space using available multi-modal data.
• Infers missing modalities by mapping source cells to target modalities.

Main Results:

• CMOT demonstrates superior performance across diverse biological applications, including developmental neuroscience, cancer, and immunology.
• The method successfully improves cell-type and cancer classifications.
• Provides robust biological interpretations from integrated multimodal data.

Conclusions:

• CMOT offers a powerful solution for multimodal single-cell data integration.
• Enhances understanding of cellular mechanisms by leveraging complementary data types.
• Advances cell-type and disease classification through accurate data fusion.