Stabilized mosaic single-cell data integration using unshared features.

Area of Science:

• Single-cell biology
• Bioinformatics
• Computational biology

Background:

• Single-cell omics technologies capture diverse biological information.
• Current data integration methods often ignore non-overlapping features, leading to information loss.
• There is a need for advanced techniques to integrate heterogeneous single-cell datasets effectively.

Purpose of the Study:

• To introduce StabMap, a novel mosaic data integration technique.
• To leverage non-overlapping features for stabilized mapping of single-cell data.
• To enable comprehensive data integration across datasets with unique feature sets.

Main Methods:

• StabMap infers a mosaic data topology using shared features.
• Cells are projected onto reference coordinates by traversing shortest paths along the topology.
• The method supports both supervised and unsupervised reference projections.

Main Results:

• StabMap demonstrates robust performance across various simulation scenarios.
• The technique facilitates 'multi-hop' mosaic data integration, even between datasets lacking shared features.
• StabMap enables the integration of spatial transcriptomic data with dissociated single-cell data.

Conclusions:

• StabMap offers a powerful approach for integrating single-cell omics data by utilizing unique features.
• The method enhances the analytical capabilities for heterogeneous and multi-modal single-cell datasets.
• StabMap advances the field of single-cell data integration, particularly for spatial transcriptomics.