Fig. 2a–c,g–i
HER2ST cohort
The explorer shows the normalized pairwise accuracy of 11 published methods for three sets of spot pairs in eight patients with HER2-positive breast cancer and ranks the methods.
Data explorers
The data explorers open the sections and saved results behind Figures 2 to 6 for browsing by gene and by sample. Each figure downloads as SVG or PNG together with the source data. Results shows all figures of the paper.
Fig. 2f
HER2ST section B1
The section has measured expression of 785 genes at 295 spots, the saved predictions of Hist2ST, EGNv2 and DeepSpaCE, and the annotated regions. Each gene can be compared over the whole section, within each region and in larger bins.
Fig. 3d, Fig. 6a
HEST sample TENX118
One lung cancer sample with 50 genes at 1,944 spots is predicted by Ridge–Phikon-v2 after standard training and with the pairwise difference loss.
Jump to
Fig. 4
Xenium breast cancer cells
Three single-cell methods are scored in five patients by RSC within malignant cells at five scales, by RSC across all cells after averaging into bins and by proliferation hotspots checked against measured RNA.
Jump to
Fig. 5a–f
External TLS sections
Two kidney and two lung sections come with expert annotations, eight TLS genes, measured and predicted, the H&E of lung section LC3, and RSC between regions and within TLS.