spSeudoMap: cell type mapping of spatial transcriptomics using unmatched single-cell RNA-seq data.

Area of Science:

• Single-cell genomics
• Spatial transcriptomics
• Computational biology

Background:

• Integrating single-cell RNA sequencing (scRNA-seq) with spatial transcriptomics is challenging due to cell type and composition mismatches, especially with sorted cells.
• Investigating specific cell populations, like immune cells, often relies on sorted scRNA-seq data, limiting direct spatial integration.

Purpose of the Study:

• To develop a computational method, spSeudoMap, for integrating sorted scRNA-seq data with spatial transcriptomic data.
• To overcome cell type and composition discrepancies between datasets for accurate spatial cell mapping.

Main Methods:

• spSeudoMap generates virtual cell mixtures from sorted scRNA-seq data to simulate spatial gene expression profiles.
• A domain adaptation model is trained to predict spatial cell compositions using these virtual mixtures.
• The method was validated on brain and breast cancer tissue datasets.

Main Results:

• spSeudoMap accurately predicted the spatial distribution and topography of cell subpopulations in tested tissues.
• The approach effectively reconciled differences between sorted scRNA-seq and spatial transcriptomic datasets.
• Successful application in complex tissue environments like brain and breast cancer.

Conclusions:

• spSeudoMap provides a robust framework for integrating sorted scRNA-seq data with spatial transcriptomics.
• This method enhances the ability to map cellular landscapes and understand the spatial roles of specific cell types.
• Facilitates deeper insights into tissue heterogeneity and the function of key cell populations.