================================================================================
SUPPLEMENTARY DATA
In Silico Identification of Natural Compound-Derived IL-18 Receptor Alpha
Antagonists as Candidate Therapeutics for Graft-versus-Host Disease
Aserrar et al.
================================================================================

CONTENTS
--------
Scripts/              all scripts used, in order of execution
Structures/           receptor and ligand structures
Vina_results/         docking output for the ten top-ranked compounds
Gromacs_results/      molecular dynamics analysis output, one folder per complex
Docking_Results_SUPDATA.xlsx   screening library, docking scores and selection


EXECUTION ORDER
---------------
1. Obabel_Conversion.sh
   Converts the ligand library from SDF to PDBQT with MMFF94 minimisation.
   Input  : Ligands_SDF/
   Output : Ligands_PDBQT/

2. VinaDocking.sh
   Runs AutoDock Vina 1.1.2 on every ligand, using config.txt for the receptor
   and the search box, then collects the best score of each ligand.
   Input  : Ligands_PDBQT/, config.txt
   Output : docking_resultats/<ligand>/out.pdbqt, docking_scores.csv

3. MD_prep.sh
   Builds the .mdp parameter files, prepares the protein-ligand system, solvates
   it, adds ions, minimises and equilibrates (NVT then NPT, 100 ps each).
   Calls MD_simul.sh at the end.

4. MD_simul.sh
   Production run, then two-step removal of periodic boundary artefacts.
   Calls MD_result.sh at the end.
   Output : *_traj_center_noPBC.xtc, the trajectory used by every analysis

5. MD_result.sh
   RMSD, RMSF, radius of gyration, hydrogen bonds, SASA, PCA, potential energy
   and free energy landscape.
   Output : the .xvg and .xpm files in Gromacs_results/

6. MD_figures_clean.py
   Regenerates every figure of the manuscript from the .xvg and .xpm files.
   Run it from the folder that contains the three complex subfolders:
       python3 MD_figures_clean.py
   Output : FIGS_CLEAN/, SVG and PNG at 300 dpi


NOTES
-----
- PREFIX is set to CNP007 in the shell scripts. Change it to CNP018 or CNP019
  to process the other two complexes.
- Group numbers passed to the GROMACS analysis tools follow the index file of
  this system: group 2 is the ligand, group 4 the protein (all atoms),
  group 7 the protein backbone.
- The production run was scheduled for 400 ns per complex. Cluster interruptions
  stopped the three runs at different times, and all trajectories were truncated
  to 340 ns, the duration reached by the shortest run. Every analysis reported
  in the manuscript covers this identical 340 ns window.
- Ligand RMSD is computed with the ligand superimposed onto its own reference
  structure. It reports internal conformational flexibility, not retention
  within the binding site, which is assessed through hydrogen bond occupancy.
- No significance test is applied to the trajectory distributions: successive
  frames are not independent observations.
- The ADMET sheet of the Excel file lists ADMETlab 2.0 values, used for the
  compound selection. The values reported in Tables 4 to 6 of the manuscript
  were predicted with ADMETlab 3.0.

Software: Open Babel 3.0.0, AutoDock Vina 1.1.2, GROMACS 2021.3 with CUDA 11.3.1,
SwissParam, PyMOL 3.1.7.2, Discovery Studio V21.0, ADMETlab 3.0.
