Umbrella sampling in OpenMM

The purpose of this project was to simulate the guest-host interactions between beta-cyclodextrins (bCD) and a ligand.

Data

The data sampled in this project is available for download at https://zenodo.org/record/7852102#.ZEJeIHZBwuU.

Requirements

Molecular dynamic simulation:
OpenMM: conda install -c conda-forge openmm Source

Openforcefield conda install -c conda-forge openmmforcefields Source

Data treatment:
FastMBAR: pip install -U FastMBAR Source
It is recommended to follow their guide to ensure a proper installation.

Molecular dynamic simulation

How to

The directory "in" contains an assorment of sdf and pdb files which should be paired up by name.

A0.py creates the initial files and directories and takes main_path as its single argument which must be constant for all the following python files (B0, C0, D0_4) A0.py copies the the nessary files from the ´in´-directory and sets up the paths at the same time. When promted either set the parameters or use the ones supplied in the benzene directory (setup_file.csv).

B0.py setup the initial pdb-files. It takes the single input main_path . The program ensures that the host and guest are centered on top of each other and the starting point for all future simulations is orego of the system. It should be suficcient to run B0.py only once per main_path

C0.py takes three arguments: main_path R0 R0_index. R0 is the distance between the host and ligand, while R0_index is the index of the distance: R0 = [0, 0.25, 0.50, 0.75, 1.00, n]; R0_index = [0, 1, 2, 3, 4, n_i]. The Umbrella Sampling is setup here with the distances of R0 as the molecules are setup with their initial distances.

D0_4.py takes the same arguments as C0.py main_path R0 R0_index. Here the host and guest is solvated in water and the production run is made. At each distance, set by R0, a CustomCentroidForce is used to keep the distance of R0. The distance is controlled by the harmonic potential $E_{pot} = \frac{1}{2} \cdot K \cdot (R-R_0)^2$, where K is the spring constant (usually set to 100 $kJ M^{-1}$ and $R$ is the measured distance.

The measured distances are written to a csv-file in main_path/production/distance/.

SLURM MD-simulation

Run the pythonfile A0.py. This will produce two additional files: C_herc.sh and D_hercules.sh. Both of these are setup for a specific slurm system but in essence, they automate much of the procedures. B0.py still has to be run seperatly.

Data treatment

This specific approach is made for data sampled through umbrella-sampling and is heavily inspired by the FastMBAR documentation.

How to

F0_3.py must be placed in the same directory as the other scripts. F0_3.py takes four arguments: main_path run_on_gpu high L_times. main_path must be the same as for D0, C0, and D0_4. run_on_gpy can be either 'yes' or 'no', high ask at what side of the bCD the ligand is placed, and L_times asks how many windows the umbrella sampling should be split into.

With the dataset suggested here, the argument high should be set to 'yes', as it is positive.

SLURM data treatment

In SLURM, run the file F0_3 and follow the same procedure as for F0_3.py.