Derived PDAC datasets and supplementary materials for "Predicting pathway activity in pancreatic ductal adenocarcinoma: linear metabolism versus non-linear immunity, and a glycolysis-stable TST anchor"

Primary cohort data were accessed from AACR Project GENIE (syn7222066, release 18.0-public) under data-use terms and are not redistributed. External validation data are available at GEO (GSE71729). This record provides derived data: GSVA scores, CIBERSORTx fractions, fixed train/test splits, model performance summaries, feature-importance tables, gene–pathway correlations, and figure source data. Changes in this version. Replaced all correlation tables and aligned figure sources to the manuscript. Included (i) updated Pearson correlation matrices for ALL/LOW/HIGH states across all genes × 15 pathways and driver_summary_PEARSON.tsv; (ii) parallel Spearman matrices (ALL/LOW/HIGH) and driver_summary_SPEARMAN.tsv for sensitivity; (iii) corrected immune GSVA header mappings (CD8 activation, IFNG signaling, NK cytotoxicity, Treg signature, macrophage activation); (iv) revised sparse heatmaps (Figure 6 all-samples; Figure 7 Δ High–Low) matching the manuscript; and (v) a Frontiers-ready figure bundle (TIFF, 18 cm, 300 dpi) plus manifest/README. A reproducible script (Generate-driver-genes-stratified.py) is included.