openmm / openmm_workshop_july2023

OpenMM: A High Performance Molecular Dynamics Library

Introduction

OpenMM is a toolkit for molecular simulation. It can be used either as a stand-alone application for running simulations, or as a library you call from your own code. It provides a combination of extreme flexibility (through custom forces and integrators), openness, and high performance (especially on recent GPUs) that make it truly unique among simulation codes.

Getting Help

Need Help? Check out the documentation and discussion forums.

openmm_workshop_july2023's Issues

section2/ US

make settings to compile uwham may not be guaranteed to work on all environments. we could document this to allow users to move on if they run into difficulties

nglview does not work in Colab

The nglview package does not work when running the notebooks in Colab. It would be great to have it working so people can quickly visualize the molecules/trajectories directly in the jupyter notebook.

Improve prerequisites

We should improve the prerequisites for this workshop material. This workshop is not intended to be an introduction to MD/Python/Jupyter. We should provide links to good suggested material on the following topics:

• Intro to Python
• How to use Jupyter Notebooks
• Introduction to molecular simulation/molecular dynamics

The CCPBioSim training materials could be appropriate for example: https://www.ccpbiosim.ac.uk/training/other-tutorials/python-for-biomolecular-modelling (Must be logged into a ccpbiosim account to view this)

tutorial 3: hybrid ML/MM trajectory visualisation artefacts

The current notebook produces a trajectory with alanine dipeptide split across PBCs. This is confusing when attempting to compare with a visualisation of a gas phase ML simulation of alanine dipeptide.
Ideally we would center alanine dipeptide in the box so that there is no time for it to drift across boundaries, or use some traj processing before visualisation.

Is it possible to modify the calls to modeller to achieve this behaviour? Alternatively can we provide a solvated topology via e.g. BioSimSpace ?

Extra figure for section_2/custom_forces.ipynb

In section_2/custom_forces.ipynb

section Create toy system

--> add a picture of the LJ system for clarity to help users understand what is being setup.

e.g.

add fix to ML/MM mixed system example

There was a bugfix in openmm-ml openmm/openmm-ml#56 that also needs to be updated in the example notebook. The example will behave correctly but only because the ML region is a contiguous set of atoms that starts from atom 0.

The table of contents links are broken

Clicking on a link in the table of content of the notebooks (one example below)

Does not lead to the relevant section of the notebook.

The links are also inactive in colab.

Restarting from checkpoint and appending StateReporter to a file results in multiple entries for the same time step.

Posting this here because it came up yesterday and I forgot to mention it, but it might be something for the main OpennMM repo.

One of the keen eyed at the workshop yesterday managed to work out that because the checkpoint file is written out every 1000 steps, there is time when a simulation finished on something that is not a multiple of 1000 - e.g. 1500.

Then restarting the simulation from the checkpoint file from step 1000 resulted in the log file having double entries.

When plotting the data, this results in two data points at that time step, although admittedly, the y values are basically identical.

I imagine it becomes more and more obvious when the disparity between the checkpoint file and the log file is greater than 10 fold and on how many steps it has to go back.

In the example below I could have got points all the way to 14900 and still restarted at 14000. Imagine if we had written a checkpoint every 10000 steps for example.

Very quick example:

 13900,-146856.87880890456,305.87961649181017,91.87334675537606
 14000,-146838.48677000694,302.0436151483807,91.85773612934197
 14100,-146570.08052000694,302.9515737651756,91.85773612934197
----------- Simulation ended, and restarted from 14000 -------------
 14100,-146569.67427000694,302.94798025386604,91.85773612934197
 14200,-146859.15565928526,305.1387575727,91.68905971263145
 14300,-147399.1003796428,305.3627325364048,91.29362283195196
 14400,-147585.3503796428,301.270411426622,91.29362283195196
 14500,-146945.1628796428,298.35001246529237,91.29362283195196

I guess a fix would be for OpenMM at somepoint in the reporter setup to check the current timestep vs the latest reported step and delete lines from the log file?

exercise 1/protein-ligand complex

In section_1/protein_ligand_complex.ipynb
Exercise 2. Download the "traj.pdb" file and visualize it.

• Provide instructions on how to do that in colab (user must click on Files icon on left hand side of UI)

• Add a screenshot of what a visualisation of traj.pdb in VMD should look like. Mention PBC visualisation artefacts.
  e.g.
  

Exercise 4. Download "bss_traj.pdb" and compare it to "traj.pdb" from method 1. What are the differences?

Is the intention to visualise both trajectories to comment on the size of the water boxes and the placement of the protein ? It's unclear how users should inspect the files (a text editor would be misleading).

Improve readability of custom force expressions

In the US tutorial

code like this

custom_nb_force = mm.CustomNonbondedForce('4*epsilon*((sigma/r)^12-(sigma/r)^6); sigma=0.5*(sigma1+sigma2); epsilon=sqrt(epsilon1*epsilon2)')

would be more readable if the string passed to the constructor is constructed step by step. e.g.

custom_lj = '4*epsilon*((sigma/r)^12-(sigma/r)^6);'
custom_lj += 'sigma=0.5*(sigma1+sigma2);'
custom_lj += 'epsilon=sqrt(epsilon1*epsilon2)'
custom_nb_force = mm.CustomNonbondedForce(custom_lj)

confusing warning on colab

When testing on colab section_1/protein_in_water.ipynb and section_1/protein_ligand_complex.ipynb

I get a warning/error that the session crashed after mamba the cell below has been executed

!mamba install -y -c conda-forge openmmforcefields

This seems fine as I can actually proceed but it may be useful to warn users that this can happen.
It may be useful to also warn that installing all the dependencies may take a few minutes.

section_3/machine_learning_potentials.ipynb

In section_3/machine_learning_potentials.ipynb

Exporting a PyTorch model for use in OpenMM

• Is it possible to add instructions to generate output at the end of the section to help the users understand what was done? maybe print properties of the torch_force ?

In the section
Simulation of alanine dipeptide with ANI-2x using OpenMM-Torch

• can the simulation of the pure ANI-2x system save a trajectory. Then ask users to visualise the trajectory.

In the section

Mixed system

• clarify that this approach only works for whole ML molecule(s)

• prompt the users to download the trajectory mixed_traj.pdb for visualisation.

= Exercises need to be introduced

= stretch goal) Is it possible to ask users to compare trajectories of a low quality MACE model vs a high quality model ? Assuming this translates into visible differences in simulated conformations of the dipeptide.