Mechanism-aware therapeutic-direction inference in β-cell-prioritized type 2 diabetes genes: reproducibility code and frozen evidence package
Abstract
Reproducibility archive for the study “Mechanism-aware human genetic evidence improves therapeutic-direction inference across β-cell-prioritized type 2 diabetes genes.” This repository contains the analysis code, frozen input manifests, prespecified mechanism panels, evidence-class definitions, derived result tables, and figure-generation scripts used in the study. The analysis evaluates therapeutic-direction recoverability across a frozen universe of 252 β-cell-prioritized type 2 diabetes candidate genes. It includes: (1) the strict pure loss-of-function/protein-truncating variant directional architecture; (2) the prespecified 20-mechanism development calibration panel; (3) the frozen mechanism-aware Tier B1/B2 architecture; (4) the source-defined 13-gene broad-evidence curation universe; (5) the 16-mechanism post-freeze external recoverability stress test; and (6) molecular-abundance, pharmacology, pathogenic-variation and safety evidence layers used as complementary translational context. Phase 5H represents the final frozen analysis. The broad-gate architecture was not modified after evaluation of the external holdout panel. Raw third-party datasets are not redistributed where redistribution is unnecessary or potentially restricted. Exact data releases, source identifiers, file hashes and retrieval information are provided in the source-data manifest. Derived tables required to reproduce the manuscript statistics and figures are included. Key frozen results include development-panel recovery of 4/20 mechanisms under the strict architecture versus 15/20 under the broad architecture, post-freeze external-panel recovery of 0/16 versus 7/16, and directional resolution of 3/252 versus 10/252 β-cell-prioritized candidate genes. Software environment: Python 3.11.Version: 1.0.0.Associated manuscript: Mechanism-aware human genetic evidence improves therapeutic-direction inference across β-cell-prioritized type 2 diabetes genes.
