Deep-Learning-Driven Optimization of Spirooxindole Inhibitors Targeting p53–MDM2 Interaction: Multiscale In-Silico Validation

Dataset Description This repository contains the computational datasets and molecular design files associated with the study "Multiscale simulation assisted discovery and optimization of spirooxindole MDM2 inhibitors" by Xuchen Xu, Kexin Cheng, Zigui Kan* (Department of Chemistry, China Pharmaceutical University). Background & Objectives The p53-MDM2 interaction is a critical target for anticancer drug development. This dataset documents a comprehensive in-silico workflow combining deep learning (Modof graph-remodeling network), pharmacophore modeling, 3D-QSAR, molecular docking, molecular dynamics (MD) simulations, and density functional theory (DFT) calculations for the optimization of spirooxindole-based MDM2 inhibitors. Contents The dataset includes: Chemical Structures & Activity Data 53 spirooxindole-based literature compounds with experimental pIC₅₀ values (training and test sets) 22 newly designed molecules (N01-N22) generated via Modof deep learning network with predicted activities and physicochemical properties (MW, cLogP, TPSA, QED, SA scores) SMILES strings and 3D optimized structures (OPLS-2005 force field) Pharmacophore & QSAR Models Ligand-based pharmacophore model ADHHR_1 (features: H-bond donor, H-bond acceptor, hydrophobic regions, aromatic ring) PLS-5 3D-QSAR model statistical metrics and contour maps Fitness scores for all compounds mapped to pharmacophore hypotheses Molecular Dynamics Simulation Data 100-ns MD trajectories (GROMACS 2019+/2021+) for MDM2 complexes with reference compound 47 and top candidates N11, N14, N15, N17 System setup: Amber99SB-ILDN (protein), GAFF (ligands), TIP3P water, 300K, 1 bar Analysis outputs: RMSD, RMSF, Rg, SASA, PCA, free energy landscapes, DCCM matrices MM-PBSA binding free energy decomposition data DFT Calculation Results Frontier molecular orbital (FMO) data: HOMO-LUMO energies, energy gaps (E_gap), ionization potential, electron affinity, electronegativity, hardness, softness, electrophilicity index Molecular electrostatic potential (MEP) maps Optimization level: B3LYP/6-311G(d) using Gaussian 16 Molecular Docking Data AutoDock Vina docking scores (affinity energies) Binding poses and interaction analyses (hydrogen bonds, hydrophobic interactions, π-π stacking) Grid box parameters: 22.5×22.5×22.5 ų, center coordinates (39.817, 11.052, 27.178) Key Findings Four candidates (N11, N14, N15, N17) showed superior binding affinities (-9.2 to -8.8 kcal/mol) compared to reference compound 47 (-8.4 kcal/mol) MM-PBSA calculations confirmed N11@MDM2 (-61.18 kJ/mol) and N15@MDM2 (-73.84 kJ/mol) as the most stable complexes DFT analysis revealed N11, N14, and N15 possess smaller HOMO-LUMO gaps (4.24-4.80 eV) indicating enhanced chemical reactivity File Formats Molecular structures: .sdf, .mol2, .pdb Trajectory files: .xtc/.trr (GROMACS) Energy data: .xvg, .csv SMILES and properties: .xlsx, .csv Figures: .png, .pdf (pharmacophore alignments, contour maps, MD analysis plots) Software Used Schrödinger Suite (PHASE, OPLS-2005), Modof (graph-remodeling network), AutoDock Vina, GROMACS, Gaussian 16, Multiwfn, Bio3D R package, PyMOL, LigPlot+ Associated Publication This dataset supports the manuscript: "Multiscale simulation assisted discovery and optimization of spirooxindole MDM2 inhibitors" (currently under submission/review). Contact For questions regarding this dataset, please contact: Xuchen Xu: [3222051390@stu.cpu.edu.cn] Kexin Cheng: [cpu_kexin_cheng@163.com] Corresponding author: Prof. Zigui Kan (cpu_zigui_kan@outlook.com) License CC BY 4.0 - This dataset is released under the Creative Commons Attribution 4.0 International license, permitting sharing and adaptation with appropriate credit to the original authors.