Unified cell segmentation for spatial transcriptomics via transcript reassignment

Tracel, a two-stage optimal transport method for spatial transcriptomics, improves cell-type separation across five human datasets, correctly assigning cell-type markers and retaining more cell-intrinsic transcripts.

• Why it matters: Spatial transcriptomics requires assigning localized molecules or bin counts to cells, but diffusion, overlap, and partial bin coverage lead to misassignment that degrades cell-level analyses.
• What they did: Tracel aligns matched single-cell RNA-seq reference cells to spatial cells to estimate expected expression at each location, then reassigns transcripts to cells by optimal transport against those profiles.
• The result: Across five human datasets spanning three platforms and three tissues, Tracel reduces neighboring-cell contamination, improves differential expression across spatial locations, and attributes ligand-receptor interactions to correct cell-type pairs.

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