mobleylab / drug-computing Goto Github PK

• The ligand-based design lecture and assignment have some inconsistencies and the code in the lecture is more helpful than the example code in the assignment, students report
• Check TeachOpenCADD content and see if any of it should be added to lectures/replace a lecture (find greatest hits/most relevant stuff for course): https://projects.volkamerlab.org/teachopencadd/talktorials.html#structural-biology
• Check content on preparing/running simulations on Google Cloud and think about incorporation (should it be added somewhere? replace something else? Does it work well?) https://github.com/pablo-arantes/making-it-rain
• Same question on these cloud-based tutorials for structural bioinformatics: https://github.com/pb3lab/ibm3202
• Same question on Simon Duerr’s MD/MC stuff: https://dev.simonduerr.eu/interactive/mdmc/intro.html
• Add discussion of “explainshell” in course background materials and probably add an example of it in first (?) lecture or wherever else it fits
• Update the shape overlay examples and assignment using an updated example from OpenEye.
• See if there's a way to migrate fortran code INTO Python so there is no need for compilation

Several notebooks do trajectory visualization with nglview; it would be helpful to migrate those to something which will work on Colab. The most important seem to be:

• MD sandbox on molecules
• Solvation sandbox/lecture

Lots of updating done in #103 , but we still need to:

• Make sure the install instructions are sufficiently clear regarding fortran libraries (in getting started materials)
• Add some recap of fortran libraries (esp for Colab) in the notebooks which require them (energy minimization, MD, MC)
• Set install part off with a separate header in the assignments

Links referring to Python_Linux_Vi in the lectures README.md are broken.

Additionally, the links in the README.md file in the Python_Linux_Vi directory are also broken.

The notebooks and worked examples for lectures need updating, particularly, these cases:

• Python, Linux, and vi(m)
• 3D structure and shape methods (#25 )
• Molecular dynamics (MD) -- currently I only have a "sandbox" for use in class and lecture material is still just slides
• Monte Carlo (MC) -- currently I have a "sandbox" for use in class and lecture material is still just slides
• Library searching -- currently I have a sandbox; rest of material is still just slides.
• Ligand-based design -- currently I have a sandbox; the rest of the material is still just slides.
• Physical property estimation
• Docking; slides have most of the material
• OpenMM, MDTraj, clustering and visualization: Probably should be updated/extended to use examples from @nathanmlim 's work on BLUES
• Mixture simulations

(I have a separate issue for the lectures which do not yet have any notebooks, #4 ).

e.g. https://github.com/MobleyLab/drug-computing/blob/b4048a06cbecced8465cd02d6755ca6d320dc504/uci-pharmsci/getting-started.md#conda-install-requirements

(Will get merged in via #1 ).

Using Toolkit version 2023.1 and presumably OEDocking TK 4.2.1.
In the notebook OEDocking.ipynb, section 3.4: I get the following error.

NameError Traceback (most recent call last)
Cell In[13], line 37
32 #Loop over 3D molecules from the input filestream
33 for mcmol in ifs.GetOEMols():
34 #for mcmol in oemols:
35 #Call our docking function
36 # Print score
---> 37 dockedMol, score = dock_molecule( dock, sdtag, num_poses, mcmol )
38 print("{} {} score = {:.4f}".format(sdtag, dockedMol.GetTitle(), score))
40 #Write docked molecules to output filestreamot defined

NameError: name 'dock_molecule' is not defined

This could possibly relate to the dock_molecule method being depreciated in a previous update as I cannot find any documentation relating to this method. In fact under the documentation for the OEDock class, it specifically states "Dock molecules using the "OEDock.DockMultiConformerMolecule"method (https://docs.eyesopen.com/toolkits/python/dockingtk/OEDockingClasses/OEDock.html#OEDocking::OEDock::DockMultiConformerMolecule).

@nathanmlim - in the getting started info (https://github.com/MobleyLab/drug-computing/blob/master/uci-pharmsci/getting-started.md) you have something like:

Verify the installation with:

python oecheminfo.py
The output should look something like:

I get a "can't open file" message when I do that. :) Probably you're missing a step about paths?

Currently, I do not have Jupyter notebooks for the following lectures, so I will need to create them (potentially from the slides), ideally also creating activities to try at the same time.

• Energy minimization
  - [ ] Quantum mechanics
• Solvation models (should probably update to have example OpenMM code using GBSA solvation, computing solvation free energy with radius updates)
• Fluctuations, correlations, and error analysis (use density example and other samples from OpenMM).
• Error analysis, part 2
• Working with proteins (whole content for this one needs replacing, not just migration from slide format, probably to use OEBio)
• Molecular visualization (needs updating to NOT exclusively use PyMol but also provide VMD intro)
• Normal modes
• Free energy calculations and the drug discovery process

Add content to solvation lecture on explicit solvent.

Please let me know if you're interested in helping with/taking ownership of content for any of these, especially if it overlaps with something you are already working on. I have slide content I could share and we could talk through what the material should cover. (cc @nathanmlim @sukanyasasmal @vtlim @bergazin )

• Improve images/make sure all have suitable credit/attribution
• Migrate to Jupyter notebook? Unsure if content lends itself to that, but would make for greater portability/easier updates.

Currently having library issues with OpenMM on Colab; this means that assignments/examples using OpenMM aren't currently working on Colab. Particularly:

• Second MD sandbox
• Solvation notebook

Jupyter notebooks for mixture simulations and host-guest binding will require updates, at least because of:

• changes in OpenFF
• issues with solvationtoolkit

The visualization lecture added a couple very large files (Keynote files) which have to managed with git-lfs. A better procedure would be to just link to them (hosted elsewhere) and remove them from this repo to make the size more manageable.

Needs updating because of issues with solvationtoolkit (still available but requires older python? Probably need to fix/make available on conda again, unless I can replace it with something from Evaluator; checking on that)

We discovered a potential issue in the order files are created in drug-computing/uci-pharmsci/lectures/solvation/solvation.ipynb. It was illustrated and discussed in SMIRNOFF99Frosst issue#71.

I think the easiest way to fix it (and what I did while testing) is to write out the pdb file at the beginning of the gas_phase_MD and GBSA_MD methods with the original oemol instead of trying to convert the openmm system back to a oemol at the end of the simulation.

Github now supports natively rendering Jupyter notebooks in the web UI. Additionally, I tried downloading one of the html files and it includes broken images. See the following example.

I can put together a PR to remove these files if the owners of this project think this is reasonable.

@nathanmlim - there is some discussion (very cursory) of virtual environments in https://github.com/MobleyLab/drug-computing/blob/master/uci-pharmsci/getting-started.md but basically it only says "while still in your virtualenvironment" but it doesn't say how to leave it or how to get back to it, which might be important.

(I don't use virtual environments that often, so another thing which also seems relevant is -- if they want to use one, do they have to open it again when they come back, or does it open by default? And what's the interplay with Jupyter notebooks, do they have to select the one they want?)

@nathanmlim - are you able to provide any info on setting environment variables that is necessary for Windows users, e.g. in https://github.com/MobleyLab/drug-computing/blob/master/uci-pharmsci/getting-started.md ?

(For example, how to get an OpenEye installation working on Windows?)

If not, I will figure this out myself.