Computational reproducibility package for MMP-9 and NLRP3 docking, enrichment, identity, and dynamics audits

<p dir="ltr">This dataset and software release supports the manuscript ‘Native-pose recovery, enrichment failure, and ligand-identity sensitivity in MMP-9 and NLRP3 docking.’ It contains frozen receptor-validation records, Vina and Vinardo retrospective benchmark datasets, corrected ligand-identity inputs and docking outputs, processed source tables from three independently seeded 100 ns MMP-9–EGCG molecular-dynamics simulations, audited MM/GBSA outputs, analysis scripts, quality-control records, figure source data and exports, provenance records, and SHA-256 integrity manifests.</p><p><br></p><p dir="ltr">Both Vina and Vinardo recovered the native poses but failed the prespecified retrospective enrichment gate.

Corrected EGCG produced a deep-sampling MMP-9 pose consensus but no reproducible NLRP3 pose; the MMP-9 simulations sampled heterogeneous Zn-proximal and alternative pocket states. These outputs support a bounded MMP-9 structural hypothesis and an audited computational workflow. They do not establish biochemical inhibition, direct NLRP3 target engagement, dual-target pharmacology, safety, pharmacokinetics, or therapeutic efficacy.</p><p><br></p><p dir="ltr">The archive includes analysis-ready data and complete file manifests but excludes the full 300 ns raw trajectory segments and bulky engine intermediates; their omission and hashes are documented in EXCLUDED_LARGE_FILES.md and the production audit.