Additional file 1 of Evaluating spatially variable gene detection methods for spatial transcriptomics data

Additional file 1: Fig. S1. Summary information and statistics of the spatial transcriptomics datasets used for evaluating concordance, statistical significance, and reproducibility of SVG detection methods in this study. Fig. S2. Pairwise correlation of SVG rankings reported by each method for individual spatial transcriptomics datasets. Fig. S3. Comparison of different gene statistics to rank SVGs. Fig. S4. (a-b) Boxplot of correlations of SVG rankings reported by each method against all other methods. Each dot denotes the results for a spatial transcriptomics dataset. The dots are coloured (a) by the total number of spatial spots in the dataset or (b) by the spatial technology platform. (c) Bar plot denoting the number of statistically significant SVGs reported by each method for each spatial transcriptomics dataset. An adjusted p-value threshold of 0.05 reported by each method for each dataset. Fig. S5. Heatmaps of the overlap of SVGs reported by each method for each spatial transcriptomics dataset. Fig. S6. Relationship between SVG statistics and proportion of zero of genes. Fig. S7. Simulation of spatial transcriptomics data. Fig. S8. Spatial patterns of spatially variant genes. Fig. S9. Spatial patterns of spatially invariant genes. Fig. S10. ROC curves of spatially variable gene detection. Fig. S11. Performance of SVGs selected by each SVG method for clustering spatial domains in the mouse embryo.