Answers are grounded in this paper and cite their sources.
This record fixes, before any new data were downloaded, the analysis protocol for an extension of a study of how much of a spatial graph encoder's agreement with tissue annotations is explained by neighbourhood averaging. The protocol defines label diagnostics of reference annotations (chance-corrected homophily across neighbourhood scale, label informativeness, a label correlation length and the predictability of a label from each observation's own expression), a construction-scale experiment, matched comparisons of three encoders (Tessera-ST, GraphST and STAGATE) with parameter-free propagation on each encoder's own graph, and the primary predictions, decision rules, margins and power limits applied to three cohorts not analysed before: mouse spinal cord MERFISH, colorectal cancer Visium and breast cancer Visium. The protocol file is identical to revision-r3/PROTOCOL-R3.md at commit 0d2326e of the study's git repository (SHA-256 e6d889dc4d3f105df06b41ddcb0a870fd4f0c2cd5e4cabf435b03cf43429e073). Its section 10 records the calibration analyses run beforehand on datasets examined in earlier work. The archive revision-r3-frozen-code.tar.gz holds the analysis code at that commit, and SHA256SUMS.txt lists both checksums.
No takes yet. Share an insight, caveat, or question.
Zeyu Fu (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: