Supplementary Material for "Glycolysis-stratified coordination of fatty acid and glutamine metabolism in pancreatic ductal adenocarcinoma""

Abstract

Update in this version: Supplementary materials were updated to align with the revised manuscript, including an updated graphical abstract and the addition of the High-versus-Low CPTAC-PDAC family-level permutation analysis (S8; Δr_meta = r_meta(High) − r_meta(Low; B=5000), with BH-FDR correction). Supplementary Material for Glycolysis-stratified coordination of fatty acid and glutamine metabolism in pancreatic ductal adenocarcinoma This repository contains the supplementary figures, tables, and source data associated with the manuscript, including predictive modeling and transcriptomic analyses (S1–S2), metabolite-inference analyses (S3–S5), fatty-acid chain-length annotations and CPTAC proteome characteristics (S6), within-stratum CPTAC proteomic analysis (S7), and High-versus-Low CPTAC family-level permutation analysis (S8). Contents: Graphical Abstract S1 – Predictive Modeling of GLN ActivityS1 Fig. Elastic-Net out-of-fold predictions for GLN activity by glycolysis grouping.S1 Table. Cross-validated performance metrics (R² and RMSE) by grouping.S1 Data. Out-of-fold predicted vs. observed GLN activity by sample. S2 – FA Family DefinitionsS2 Table. Curated mapping of fatty-acid (FA) pathways/enzymes to functional families used in transcriptomic analyses. S3 – Inferred Metabolite ActivitiesS3 Fig. Principal component analysis (PCA) of inferred metabolite activities across PDAC samples.S3 Table. Raw and z-scored inferred metabolite activity scores for 13 metabolites. S4 – Metabolite State ComparisonsS4 Fig. Concordance heatmap between glycolysis strata (Low/Medium/High) and metabolite-defined tumor states (sample counts).S4 Table. One-way ANOVA and Kruskal–Wallis statistics comparing metabolite activities across phenotypes. S5 – Mapping and Final State LabelsS5 Table. Metabolite–gene mapping with +1/−1 directionality assignments used to compute weighted metabolite activity scores.S5 Data. Final metabolite-defined metabolic state labels for all PDAC samples. S6 – FA Chain-length Schematic and CPTAC Proteome CharacteristicsS6 Fig. Schematic mapping representative enzymes and pathways across fatty-acid chain-length classes (MCFA, LCFA, VLCFA) and two example lipid families (glycerophospholipids, sphingolipids).S6 Table. Descriptive statistics for the CPTAC PDAC tumor proteome (MD_abundance_tumor matrix), including number of tumors (n=140), total proteins (11,662), per-protein missingness distribution, and the number of proteins retained after filtering to ≥70% observed values (7,993). S7 – Proteomic Validation of FA–GLN Coordination (CPTAC-PDAC Cohort)S7 Fig. Proteomic validation of sphingolipid–glutamine coordination by glycolysis state in PDAC. Family-level associations between GLN and FA protein modules (glycerophospholipid and sphingolipid) are shown across glycolysis strata.S7 Table. Proteomic validation statistics for FA–GLN family-level correlations across glycolysis strata in the CPTAC-PDAC cohort (n=140), including sample size, Spearman correlation between GLN and FA protein-module scores, nominal P values, and Benjamini–Hochberg FDR-adjusted q values. S8 – High-versus-Low CPTAC-PDAC ValidationS8 Table. Δr_meta = r_meta(High) − r_meta(Low), with permutation P values (B=5000) and Benjamini–Hochberg FDR-adjusted q values.S8 Method. Permutation-test documentation for the High-versus-Low analysis (B=5000). Contact: Correspondence available upon request.