Baseline immune transcriptional signatures are prognostic but do not mediate the TMB-survival association under PD-(L)1 blockade

This deposit contains the complete analytical pipeline, harmonised input data, and per-step result tables supporting the re-evaluation of combined-biomarker assumptions in the tumour mutational burden (TMB)-survival association under PD-(L)1 blockade. The analysis re-uses five published, de-identified checkpoint-blockade cohorts in urothelial carcinoma, melanoma, and non-small cell lung cancer (NSCLC), with 499 samples entering at least one analytical layer, together with one single-cell deconvolution reference (Chen 2020) and one single-cell audit dataset (Sade-Feldman 2018). The pipeline reproduces the full manuscript inferential chain: - Lange-Hansen inverse-probability-weighted Cox mediation decomposition with closed-form linear-Gaussian mediator-density ratios- Product-of-coefficients Cox cross-engine validation (VanderWeele 2011)- Cross-cohort Hartung-Knapp-Sidik-Jonkman (HKSJ) random-effects pooling of the exposure-to-mediator coefficient- Cell-type deconvolution of the IMvigor210 bulk transcriptome onto a Chen 2020-derived single-cell signature matrix- Sade-Feldman 2018 single-cell audit: Harmony integration, Leiden clustering, ARI assessment against TISCH lineage labels, and Hungarian-algorithm cluster alignment- SU2C-MARK NSCLC supportive-concordance analysis under the decomposed-arrows framework- Sensitivity battery: restricted cubic spline βA, CD274/PD-L1 orthogonality, linearisation-residual and effective-dimension diagnostics, additive-hazard (Aalen) mediation cross-check, Schoenfeld proportional-hazards diagnostics- Supplementary robustness: detection-and-materiality envelope, Bayesian algebraic triangulation- Post hoc calibration: synthetic-mediator engine calibration and enrichment-rule simulation All numerical results and tables reported in the manuscript are reproducible from these scripts given the deposited harmonised inputs and the fixed random seeds documented in Supplementary Section S1.15.