Help us secure federal funding for OpenMM!

/

Recently, OpenMM applied for NIH funding, and while we just missed being funded this round, we’re optimistic about being funded in our resubmission. Besides powering a large fraction of the biomolecular simulation community (OpenMM Has been downloaded over 380,000 times from conda alone), we’re super excited to be working be able to tightly integrate OpenMM with modern ML frameworks such as TensorFlow, PyTorch, and JAX. We are building OpenMM plugins to allow machine learning potentials or collective variables defined in these machine learning frameworks to be easily used within OpenMM, and Python libraries to make it easy to build next-generation hybrid quantum machine learning / molecular mechanics (QML/MM) models within these frameworks. Our earliest results already show that hybrid QML/MM simulations where the ligand is treated with QML can drastically reduce the error in alchemical protein-ligand binding free energy calculations from 1 kcal/mol to 0.5 kcal/mol! We’ll also build a common model repository that will allow popular QML models to easily be used in your simulations. You can read more about our plans in our research proposal, posted here.

EDIT: The submitted research proposal is available here. Thanks to all of you who submitted letters of support!

We need your help!

If you either currently use OpenMM in your research or software, or would like to in the future, please consider writing us a Letter of Support for our NIH resubmission! All you have to do is draft a letter on institutional letterhead, addressed to

Tom Markland
Associate Professor, Department of Chemistry
Stanford University

and cover any or all of the following bullet points:

  • How you currently use OpenMM in your research or software, or how you plan to use it

  • How additional support for quantum machine learning potentials, machine learning collective variables defined in PyTorch/TensorFlow/JAX, or complex integrators defined in these packages will be useful to you

  • Why you think machine learning potentials or collective variables are going to be useful for biomolecular simulation

  • How you would be able to make use of our specific proposed work in the grant proposal, such as support for advanced potential functions, continued hardware optimizations, machine learning framework plugin support, hybrid QML/MM potentials, accelerated physical MM force Op library for machine learning frameworks, or the generation of large quantum chemical datasets for biomolecules on Folding@home that will be deposited real-time into the MolSSI QCArchive quantum chemistry archive

Finally, send a PDF copy of the Letter of Support to openmm@choderalab.org by Friday 23 October.

We also welcome your feedback on how OpenMM can continue to serve the needs of the biomolecular simulation community over the next decade and beyond!

Thanks so much!

John Chodera
Tom Markland
Gianni de Fabritiis

COVID Moonshot seeks NIH funding

/

The COVID Moonshot—our patent-free, open science effort to discover an orally available inhibitor for SARS-CoV-2 main viral protease that could be used for treatment or prophylaxis following exposure—submitted a proposal to the NIH NIAID COVID-19 Emergency R01 program for funding to complete our task to deliver an inhibitor for IND-enabling studies! Award decisions should be made in Sep 2020, and funds beginning Oct 2020.

You can read the scientific component of the proposal (submitted 2020-08-14) here: [PDF]

UPDATE: NIH timelines have pushed back proposal review to Jan 2021.

UPDATE: We submitted a one-page supplement with additional preliminary data on 2020-12-14: [PDF]

UPDATE: Summary Statements have been made available on 2021-02-01: [PDF]

Securing sustainable funding to enable OpenMM to continue to power the next decade in biomolecular modeling and simulation

/

OpenMM is the most widely-used open source GPU-accelerated framework for biomolecular modeling and simulation. It has been cited more than 1300 times, downloaded over 280,000 times from conda-forge alone, and has run on more than one million distinct computers. Its Python API makes it widely popular as both an application (for modelers) and a library (for developers), while its C/C++/Fortran bindings enable major legacy simulation packages to use OpenMM to provide high performance on modern hardware. OpenMM has been used for probing biological questions that leverage the $16B global investment in structural data from the PDB at multiple scales, from detailed studies of single disease proteins to superfamily-wide modeling studies and large-scale drug development efforts in industry and academia.

Originally developed with NIH funding by the Pande lab at Stanford, we now aim to fully transition toward a community governance and sustainable development model and extend its capabilities to ensure OpenMM can power the next decade of biomolecular research, guided by the OpenMM Consortium. To fully exploit the revolution in QM-level accuracy with quantum machine-learning (QML) potentials, we also plan to add plug-in support for QML models augmented by GPU-accelerated kernels, enabling transformative science with QM-level accuracy. To enable high-productivity development of new ML models with training dataset sizes approaching 100 million molecules, we will develop a Python framework to enable OpenMM to be easily used within modern ML frameworks such as TensorFlow and PyTorch. Together with continued optimizations to exploit inexpensive GPUs, these advances will power a transformation within biomolecular modeling and simulation, much as deep learning has transformed computer vision.

Recently, we applied for federal funding to realize this vision via a new NIH Focused Technology Research & Development R01 proposal, with strong support from the biomolecular simulation community. You can read the scientific components of the proposal we submitted here: [PDF]

We welcome your feedback on how OpenMM can continue to serve the needs of the biomolecular simulation community over the next decade and beyond!

Slides from Labsgiving group meeting on software development best practices

/

It’s become a tradition in the lab to have a Labsgiving group meeting the week of Thanksgiving! This year, I was on the group meeting rotation, and we were lucky enough to have our wonderful collaborators from the Andrea Volkamer lab at the Charité in Berlin join as well!

PANO_20191126_103954.vr (1).jpg

I covered some topics on why we practice open science and open source software, and our current best practices for doing so. Best practices are ever evolving, and we are always learning ways to better practice our craft and share our work with the world. In that spirit, we have tried to capture much of our current practices in a GitHub repo to make it easier to gather feedback and continually refine our practices: [software development best practices]

I put together a short talk covering some of the highlights of our best practices. In case it’s useful, you can find the PDF slides here: [PDF]

Happy Thanksgiving, all!

~ John

Slides from John Chodera's talk at GCC 2019

/

I was thrilled to have had the opportunity to speak at the 15th German Conference on Cheminformatics (GCC 2019) in Mainz! You can find a PDF version of my slides here: [PDF] [reduced size PDF]

I’m also tremendously excited that GCC 2020 will take place 1-5 Nov 2020 and will feature two days of EuroSAMPL, our first international SAMPL challenges meeting! Stay tuned for more information.

The Chodera lab teams up with Andrea Volkamer to explore the interface of machine learning and free energy calculations

/
Professor Andrea Volkamer, Charité (Berlin) works at the frontier of structure-guided machine learning for drug discovery and kinase inhibition.

Professor Andrea Volkamer, Charité (Berlin) works at the frontier of structure-guided machine learning for drug discovery and kinase inhibition.

The Einstein Foundation and Stiftung Charité have awarded a BIH Einstein Visiting Fellowship to a new collaboration between the Chodera lab and Professor Andrea Volkamer of the Charité in Berlin to develop new approaches that meld structure-informed machine learning with free energy calculations to predict and design kinase polypharmacology. We will be hiring a postdoc and PhD student to be embedded within the Volkamer group within the exciting human health research environment at the Charité in Berlin. John Chodera will be making four extended visits to Berlin each year, and a significant travel budget will allow research personnel to make extended visits to both the Volkamer group at the Charité and the Chodera lab at the Memorial Sloan Kettering Cancer Center (MSKCC). [Job Postings]

An example workflow for utilizing structure-informed machine learning and free energy calculations to predict kinase polypharmacology.

An example workflow for utilizing structure-informed machine learning and free energy calculations to predict kinase polypharmacology.

Celebrating six years of the Chodera lab!

/

The Chodera lab turns six years old on 1 Nov 2018! It’s been an exciting six years, having submitted or published thirty-five papers and having twenty-four trainees pass into or through the lab since then. To celebrate, John Chodera was asked to give a Sloan Kettering Institute (SKI) talk highlighting some of the exciting accomplishments and future directions. You can find the video of the whole talk here: [YouTube] [PDF Slides]

Check out more videos at the Chodera lab YouTube Channel.

Congratulations to TPCB graduate student Chaya Stern for being selected for a 2018 MolSSI Phase I Fellowship!

/

Congratulations to Tri-Institutional PhD Program in Chemical Biology (TPCB) graduate student and NSF GRFP recipient Chaya Stern on receiving a 2018 MolSSI Phase I Fellowship to support her work in developing new algorithms and open source software for Bayesian inference of force field parameters from experimental and quantum chemical data! You can learn more about Chaya's work in this area by listening to her PyData NYC 2017 talk or reading her MolSSI Fellowship Proposal, and hear more about what Chaya is up to by following her twitter feed.

2018 Workshop on Free Energy Methods, Kinetics and Markov State Models in Drug Design: Videos are up!

/
drugmsm.jpg
drugalchemy.jpg

The 2018 Workshop on Free Energy Methods, Kinetics and Markov State Models in Drug Design held at the Novartis NIBR campus in Cambridge, MA was a huge success! You can read some of the livetweeting by following #drugalchemy and #drugmsm on twitter, and catch up on talks you may have missed by going through our YouTube video playlist featuring recordings of all talks where the speakers consented to being recorded.

Special thanks to lead organizers Michael Schnieders and Greg Bowman, the fantastic panel of organizers, postdoc Levi Naden and Novartis Investigator Callum Dickson for wrangling A/V and venue details, Jose Duca for graciously allowing us to use the NIBR site, the Novartis A/V team, keynote speaker Mark Murcko, our hosts at Novartis, and all our wonderful speakers and participants for making this a success.

Seeking a joint Open Force Field Consortium / XtalPi Distinguished Postdoctoral Fellow

/

The Open Force Field Consortium (OpenFF, openforcefield.org) and XtalPi, Inc. (xtalpi.com) seeks a Distinguished Postdoctoral Fellow to perform cutting-edge research in a unique academic-industrial joint effort.

OpenFF is an academic collaboration (based at UC Irvine / UC Davis / UC San Diego / Univ. Colorado Boulder / Sloan Kettering Institute in New York) that seeks to develop next-generation molecular mechanics force fields and associated parameterization software and data infrastructure. XtalPi is a pharmaceutical technology company founded in 2014 that is introducing revolutionary advances in drug research and development, and has established strategic partnerships with several major pharmaceutical companies and recently completed a Series B funding round with Sequoia, Tencent, and Google.

The Postdoctoral Fellow will work with OpenFF and XtalPi to improve the accuracy of molecular mechanics force fields for predicting crystal structures and binding free energies. The work spans the disciplines of force field development/validation and Bayesian inference, and seeks to answer the fundamental scientific questions: 1) What regions of chemical space are critical failures for current force fields? 2) What are the fundamental limitations of (a) current functional forms and (b) parameterization methods?

The position spans a two-year project period that involves a gradual transition from the OpenFF side to the XtalPi side. In Year 1, the Fellow will be hosted for ~9 months in one of the academic groups (to be determined by the OpenFF PIs, XtalPi, and the Fellow) and ~3 months at XtalPi in Shenzhen, China or Boston, MA. In Year 2, the time division will be 3-6 months academic / 6-9 months XtalPi. The Fellow will be considered for full-time employment at XtalPi after Year 2.

For more information, see the full job posting.

The Open Forcefield Consortium seeks a software scientist

/

The Open Forcefield Consortium [http://openforcefield.org] seeks a Lead Software Scientist to coordinate open source software development efforts for an interdisciplinary academic team developing next-generation molecular mechanics forcefields and associated parameterization infrastructure.

See the full job ad here.

Postdoc Sonya Hanson joins Nobel Laureate Joachim Frank's lab

/

Congratulations to Postdoctoral Fellow Dr. Sonya Hanson, who has joined the laboratory of Joachim Frank at Columbia University to learn the exciting new technique of cryo-EM spectroscopy coupled with manifold embedding to reveal the dynamic conformational landscapes of TRP channels, an important class of integrative sensing proteins Sonya extensively studied as a graduate student. Frank was recently awarded the Nobel Prize in Chemistry (along with Jacques Dubochet and Richard Henderson) for his seminal contributions to advancing the technology of cryo-electron microscopy to become a powerful technique for imaging the conformations of biomolecules. Sonya will maintain strong links with the laboratory and the Folding@home Consortium as she progresses toward an independent faculty position.

You can see more fantastic work from Dr. Hanson at her Google Scholar page, check out her recent paper on the heat activation of TRPV1 in PNAS, and find out more about her career trajectory at her website.

Postdoc Gregory Ross joins Schrödinger as a Senior Scientist

/

We're excited to announce that Postdoctoral Fellow Dr. Gregory Ross has joined Schrödinger as a Senior Scientist, where he will be working to bring his expertise in statistical mechanics and semigrand canonical methods to their suite of molecular modeling and simulation tools.

You can see more fantastic work from Dr. Ross at his Google Scholar page, and check out his recent preprint on semigrand canonical methods for simulating realistic biomolecular salt concentrations on bioRxiv.