Supplementary Material for "Accelerating Fragment Baed Drug Discovery using Grand Canonical Nonequilibrium Candidate Monte Carlo"

This repository contains the data and scripts necessary to reproduce the results presented in our manuscript, "Accelerating Fragment Baed Drug Discovery using Grand Canonical Nonequilibrium Candidate Monte Carlo", by W. G. Poole, M. L. Samways, D. Branduardi, R. D. Taylor, M. L. Verdonk, and J. W. Essex (2024).

There are separate directories for the bulk concentrations simulations as well as the three test systems: beta-cyclodextrin, T4L99A, and MUP1 with the scripts and input files required to run the GCNCMC/MD simulations. Where the necessary scripts to reproduce the published data can be found. The required analysis can also be found in each sub-directory. More information on how to perform the simulations and analysis can be found within each directory. The grandlig module required to run these simulations can be downloaded here. Installation instructions are provided on the main repo.

If you encounter any problems with these scripts, please open an issue or contact the authors.

Overview

Summary

We have providied to avenues to reproduce the results of the paper. The simpler approach is to use the scripts in Basic_Simulation with different arguments to recreate the results. This will require the user to set up their own folder structures etc etc.

The second approach can be to navigate through the directories HostGuest, T4L99A, MUP1 where there are set up folder structures with individual python scripts / information in each which can be run to generate results. A lot of simulations have been performed for this paper and therefore regenerating these results is no small task and will require the use of HPCs.

Finally, while best efforts have been taken to ensure everything in this repo runs and works as expected, there are clearly many scripts and input files provided. If you run into any issues please open an issue and we will amend them ASAP.

Manifest

Concentration

Input files to run the concentration simulations.

Ligands

Text files containing the name, smile string and calculated excess chemical potentials for the ligands studied in this paper. A notebook is provided to recreate the figures in the SI depicting the structure of the molecules.

Mu_Ex_Calculation

Example scripts and input files to calculate the excess chemical potential for some of the ligands studied.

HostGuest

Scripts and input files to reproduce results in the paper regarding the host guest beta-cyclodextrin system.

T4L99A

Scripts and input files to reproduce results in the paper regarding the T4L99A system.

MUP1

Scripts and input files to reproduce results in the paper regarding the MUP1 system.

Basic_Simulation

Basic scripts which can be used to run sphere based GCNCMC simulations. These scripts use argparse to allow the used to provide all the relevent information to run a simulation. While input files and run scripts are provided in the above sections, these scripts can also be used to reproduce the results of the paper. This route provides more flexibiliy may be easier to work with.

Analysis_Scripts

Set of scripts required to perform various pieces of analysis. Again, these can be manipulated to used as inspiration for your own analysis and beyond.

Outputs

Collated output data which matches the publication.

Generalisation

This repo has many many input files. This is so that the results in the paper are reproducible. To make things simpler and clearer this is a more general overview to the GCNCMC protocols.

For any given ligand and system:

1. An excess chemical potential needs to be calculated for the ligand. This is basic hydration/solvation free energy calculation. Note that as mentioned in the paper, for suffciently dilute concentrations and hydration free energy will suffice. For higher concentrations, depending on application, it may be more precise to perform a solvation free energy calculation in a solution of that concentration. Examples are found in T4L99A/FEs/Solv, the main repo, and in Mu_Ex_Calculations.

2. Now the user must make choices on what kind of simulation they want to run. To find occluded sites you may choose to define a sphere which covers the whole protein, and run at a certain concentration. I.e. T4L99A/BindingSite. Or to calculate free energies via titration you may want to set the sphere to cover the binding site and run simulations at a range of cocnentraions/B values. I.e. T4L99A/Titrations