Data and code for: Molecular Dynamics and AI-Assisted Ligand Generation Identify a Cryptic Binding Site Candidate in CTX-M-15 beta-lactamase

This dataset supports the manuscript “Molecular Dynamics and AI-Assisted Ligand Generation Identify a Cryptic Binding Site Candidate in CTX-M-15 beta-lactamase.” The archive contains data and code from a computational workflow used to identify and evaluate a ligand-stabilized adjacent pocket (LSAP) candidate in CTX-M-15 beta-lactamase. The workflow included replicate molecular dynamics simulations of apo and avibactam-bound CTX-M-15, fpocket-based ensemble pocket detection, Pocket2Mol ligand generation, short protein-ligand molecular dynamics triage, and final triplicate 200 ns simulations of four LSAP-centered candidate molecules. The archive includes raw production output files for the six initial apo and avibactam-bound holoenzyme 200 ns simulations, fpocket screening outputs, selected pocket files, Pocket2Mol input and output files from the first-round and LSAP-centered generation stages, source files and summary outputs for 20 ns triage simulations, processed/fitted 200 ns trajectories for the four final LSAP-centered candidates, representative protein-ligand structures, analysis scripts, tabulated source data, file manifests, and SHA256 checksums. The data are intended to support reproduction of the reported manuscript figures, tables, screening-funnel summaries, protein-stability analyses, ligand-to-LSAP distance analyses, and final LSAP-associated fraction calculations. These files document computational discovery and prioritization only; they do not constitute experimental evidence of ligand binding affinity, enzymatic inhibition, or an allosteric mechanism.