Reproduction of Empirical pval in Visium mouse kidney datasets

[pepsiv]

Hi! Thanks a lot for developing such a great method—it’s been really valuable.

I’m trying to reproduce the Visium mouse kidney (AKI vs control) results for spatialCorrelationGeneExp, but I’m running into two issues: (1) the results aren’t reproducible across runs (the empirical p-values change each time), and (2) for most genes, my empirical p-values don’t match the values reported in your Visium kidney example. I followed the Methods exactly for preprocessing (CPM/library-normalized input, ≥1% gene filter, B=100 permutations, δ grid including 0.01 and 0.05, matched/shared spot IDs + coordinates).

Would you be able to share the exact data + script you used for the Visium kidney empirical p-values? I’d like to pinpoint where my setup differs and understand what’s causing the discrepancy.