openmm / openmm-org Goto Github PK

We may want to list and link to

• Current binary releases
• Older binary releases
• Conda installation directions

The OpenMM User's Manual, Python API docs, C++ API docs, and Developer Guide are all served over HTTP rather than HTTPS. We should serve via HTTPS and redirect HTTP requests to HTTPS.

We need to update the listed developers/contributors

Hi,

Thanks for making this great tool.
Is there anywhere I can find example or tutorial for Protein-Ligand Complex modeling?
I'm thinking of something like GROMACS tutorial of the same subject.

Thanks and hope to hear from you again .

G.V.

I pushed some changes to the master branch yesterday, but they still don't seem to be showing up in the live website. Is there something else I need to do?

Reminder issue per call today to allow at least @jchodera write access to this repo, and possibly me (@Lnaden) to make managing the site less burdensome on a few people. Also allows easier example additions.

@peastman: Should we add the members of the OpenMM Consortium to the website?

When I submitted the request to google to re-index the website, it notified me of these potential usability issues:

Hi, could we update the website to provide the following information at https://openmm.org/development

• OpenMM team

Michel lab
Julien Michel
Stephen Farr

• Funding Sources

EPSRC
OpenMM is currently (November 2022 - October 2025) supported by the Engineering and Physical Sciences Researh Council under grant EP/W030276/1 .

With the latest updates to the benchmarks, the table has started looking kind of ugly, because the column hearers are broken in strange places. At least on my computer, for example, I see

CUDA (GTX Titan
X)

There's no reason it needs the line break. There's plenty of room for the table to become wider. But it isn't doing it.

We need to update the benchmarks page with numbers for 8.0 and for newer GPUs. To start, here are the numbers for RTX 4080. I ran them with the command

python benchmark.py --platform=CUDA --style=table --test=pme,apoa1pme,amber20-cellulose,amber20-stmv,amoebapme

I repeated it three times. Here are the speeds, along with the average over the three runs.

In all cases the variations around the average are a few percent or less.

The openmm.org certificate has expired.

@peastman : Do you know anything about how SSL certificates work, or how to renew ours?

I have been trying to install the openmmforcefields conda package following the instructions on http://docs.openmm.org/latest/userguide/application/02_running_sims.html#force-fields (conda install -c conda-forge openmmforcefields). However I keep getting the same result shown in the picture below. Any suggestions? I have anaconda installed on Windows 10.

.

Wanted to make an issue before making a PR updating the tutorials to more closely follow PEP8 and use modern features of the python language. Since python3.6 is the oldest version of python supported in the next openmm release (at least that is my understanding) I would not use any features newer than those available in 3.6.

My motivation is that most users will copy and paste a tutorial and adapt things to fit their workflow, so we should try and incorporate python programming best practices.

Two quick examples from the alchemical free energy tutorial.

# Defining alchemically-modified potentials
sigma = 3.4*unit.angstrom; epsilon = 0.238 * unit.kilocalories_per_mole
# Use a comma to define related things close together 
sigma, epsilon = 3.4*unit.angstrom,  0.238 * unit.kilocalories_per_mole

# Simulating alchemically-modified systems
print('state %5d iteration %5d / %5d' % (k, iteration, niterations))
# use f-string formatting, print the current state number, and take care of starting at 0 and not 1
print(f"state {k} / {nstates-1} iteration {iteration} / {niterations-1}")

I'm happy to submit a PR for each tutorial and we can discuss each change on a case-by-case basis.

I have prepared some Jupyter notebooks to help MD practitioners new to OpenMM get started:
https://github.com/choderalab/openmm-tutorials

It would be great to include these in the OpenMM Tutorials page.

One approach is to use jupyter nbconvert to convert them to Markdown, but this loses a lot of formatting of the output cells and graphical displays.

An alternative might be to add jekyll-jupyter-notebook support to the Jekyll static site renderer and just include the notebooks directly.

I just noticed that the benchmarks seem to be missing. When I go to the Benchmarks tab it describes all the benchmarks we ran, but doesn't list any of the results. I'm pretty sure they used to be there. Something seems to have broken it.

What do you think about adding a page to make it easy to cite OpenMM, like MDTraj does?

The dihydrofolate reductase (DHFR) models benchmark (Joint Amber/Charmm benchmark) link on
https://openmm.org/benchmarks
gives a page not found -404 error. (http://ambermd.org/amber10.bench1.html)

The navigation bar is pretty much useless on mobile:

Can we turn that bar into a hamburger (three vertical lines showing a drop-down navigation menu) on mobile?

In preparation to the new website (#45), we need to move to the GitHub Pages (https://docs.github.com/en/pages). Apparently, even the current (old) website works fine on the GitHub Pages: https://openmm.github.io/openmm-org/

To complete the migration, the DNS record of openmm.org has to be update: https://docs.github.com/en/pages/configuring-a-custom-domain-for-your-github-pages-site/managing-a-custom-domain-for-your-github-pages-site#configuring-an-apex-domain

As a bonus, this will solve the HTTPS issue (#38).

The new layout has an awkward whitespace-to-text ratio, with the fonts being plain and tiny with way too much whitespace between them:

Could we adjust the font sizes for better whitespace-to-text ratios, and maybe adjust the font family to be a bit more modern?

Greetings,
when I try to open the openmm website using Edge, Chrome, or Internet Explorer browsers, a message appears like in the attached image.

is this an error with my browsers?

I think that this is not really representing the real situation. I suggest to change Gromacs by ACEMD.

Previously, docs.openmm.org redirected to the documentation.

It still redirects to https://openmm.org/documentation.html, but this no longer points to the documentation page.

Fixing the redirect from the DNS server won't work because there is no way to specify a URL that goes directly to the documentation tab. This seems problematic.

How can we fix this? Can we add some way to have a URL point directly to the documentation tab?

Should we list the latest major.minor version number on the webpage?

Maybe "OpenMM 7.4" instead?

We could also add a link to "See latest release notes" so folks could check out what is new in the latest version, linking to the GitHub releases page:
https://github.com/openmm/openmm/releases

It appears I have write access, but not admin access, so @peastman will have to lock the master branch from direct commits.

• The current performance benchmarks (https://openmm.org/benchmarks) are just for OpenMM 7.4 and missing the latest GPUs (Turing and Apere cards).
• Only DHFR is on the OpenMM website, while the rest are on the SimTK website (https://simtk.org/plugins/moinmoin/openmm/Benchmarks).

How is the http://docs.openmm.org redirect to http://openmm.org/documentation.html work, and who has control over that?

I've created a preview branch we should make changes to (or create PRs against). This will push the preview to http://openmm.org/preview/. We should create an http://preview.openmm.org redirect.

I've been using OpenMM as the main MD engine inside of wepy for a few years now and it would be nice to have it added to the ecosystem page.
Let me know what else you want for it.

Here is the description:

A Weighted Ensemble (WE) simulation framework in Python with first class support for OpenMM.

@jchodera and company are working on examples / tutorials to support the openmm paper. Per the openmm call last Friday, we want them on the openmm website. I think @jchodera has them in markdown format.

I'm going to add the jekyll static site generator (used via github pages) so we can support markdown documents. I'll also refactor out the boilderplate html (styling, menus). The current pages will stay as html

@jchodera : do you have drafts of the documents that I can use as placeholders?

The code i've been using for the last three months worked well until Monday, but now it appears this message:
python3 code.py
Traceback (most recent call last):
File "/home/mateoc/Documentos/practiques/tleap_propka/code.py", line 1, in
from simtk.openmm.app import *
File "/home/mateoc/miniconda3/envs/openmm/lib/python3.10/site-packages/simtk/init.py", line 1, in
import openmm
File "/home/mateoc/Documentos/practiques/tleap_propka/openmm.py", line 1, in
from openmm.app import *
ModuleNotFoundError: No module named 'openmm.app'; 'openmm' is not a package

the way I used to import this libraries is this one:
from openmm.app import *
from openmm import *
from simtk.unit import *
from openmm import app

I think @mpharrigan has access to these buckets, but given that he is moving on, we should make sure others have access.

@peastman : Do you have access as well? Can you designate someone else with access too?

@giadefa just made me the admin of this repository. So, I'll dedicate some of my time maintaining and improving http://openmm.org

@jchodera @peastman First, I want to discuss what would be priorities. Are any specific issues I should start with?

In general, my opinion is that the navigation is a bit messy. There are 3 buttons at the top, 5 big links an the bottom, this should be organized into a single menu structure. Also, the home page is quite empty, probably we could add the code spinet from http://openmm.org/about.html and one of the embedded videos from http://openmm.org/documentation.html

Now that we are NIH-supported, we should add an acknowledgment to our NIH support and NIH grant R01GM140090.

There seems to be significant variation in the performance of different models/variants of the same GPU (e.g. the multiple variants of A100 available), so we should provide more details in our benchmarks about exactly which model(s) were used.

It would be great to have some information on what hardware is required to run OpenMM