AI-predicted spatial transcriptomics unlocks breast cancer biomarkers from pathology.

AI model "Path2Space" predicts spatial gene expression from histopathology slides, accurately inferring thousands of genes and outperforming 21 established methods in breast cancer.

• Why it matters: High costs of spatial transcriptomics limit large-scale biomarker discovery, creating a need for affordable, scalable alternatives to molecular assays in cancer research and treatment.
• What they did: Researchers trained "Path2Space" on extensive breast cancer spatial transcriptomics data to predict gene expression and tumor microenvironment features in 976 TCGA tumors, identifying subgroups with distinct survival outcomes.
• The result: The low-cost spatial TME landscapes derived by "Path2Space" enable more accurate predictions of patient response to chemotherapy and trastuzumab, facilitating large cohort biomarker discovery and tumor biology insights.