Enhancing cell type deconvolution in spatial transcriptomics through incorporating multiple reference panels.

SWIMR, a multi-reference spatial transcriptomics deconvolution framework, recovered ground-truth cell-type compositions with highest accuracy in human heart tissue and improved resolution of compartments elsewhere.

• Why it matters: Single scRNA-seq references can be incomplete or mismatched, limiting deconvolution accuracy in complex tissues where each spatial spot contains multiple cell types.
• What they did: SWIMR integrates multiple scRNA-seq reference panels by modeling each spatial location as a weighted combination of references, with both cell-type proportions and reference weights varying across tissue while accounting for spatial correlations.
• The result: Across comprehensive simulations and diverse ST datasets, SWIMR consistently outperformed single-reference approaches, delineating tumor, stromal, and immune compartments in human lung and breast cancer and resolving organ-specific cell types in mouse embryo and liver cancer.

AI summary of the abstract · may contain errors · not medical advice · read the paper.