COME: contrastive mapping learning for spatial reconstruction of single-cell RNA sequencing data.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) provides high-throughput transcriptomic data at single-cell resolution.
• Spatial information is critical for understanding cellular functions and disease mechanisms but is often lost during tissue dissociation.
• Spatial transcriptomic (ST) technologies offer spatial gene expression data but face limitations in gene assay capacity, cost, and cell-type annotation granularity.

Purpose of the Study:

• To introduce COME, a Contrastive Mapping lEarning approach for recovering spatial information in scRNA-seq data.
• To enable precise cell-spot correspondence learning between scRNA-seq and ST datasets.
• To leverage existing scRNA-seq data to infer spatial properties and enhance biological insights.

Main Methods:

• Developed COME, a contrastive learning framework for mapping scRNA-seq data to ST data.
• Utilized cell correspondence learning to transfer spatial information.
• Validated COME through extensive experiments comparing its performance against existing methods.

Main Results:

• COME effectively captures precise cell-spot relationships, outperforming previous methods in spatial recovery for scRNA-seq data.
• The method accurately identifies biologically relevant spatial information, including missing gene structures and hierarchical patterns.
• COME precisely determines cell-type compositions for individual spots within the spatial context.

Conclusions:

• COME successfully recovers lost spatial information from scRNA-seq data by integrating with ST data.
• The approach provides valuable insights into tissue heterogeneity and cellular activities within their spatial microenvironments.
• COME enhances the understanding of complex biological systems by integrating transcriptomic and spatial data.