OMSF Newsletter - Winter 2025
Happy New Year!
As winter settles in for most of us, we are happy to share the latest updates from our ecosystem, including new hires, future announcements, and some incredible software releases. Cozy up with us, and let's reflect on the last of 2025!
Community News
We've welcomed a new face to our ever-growing OMSF staff - Jan Domanski has joined the OpenFold team! Coming from directly from CHARMM Therapeutics, Jan brings extensive experience in molecular modeling and simulation to OpenFold as the Senior Technical Lead. He's already contributing to the ongoing development of OpenFold 3, and we are excited to have him on board. Welcome Jan!
We previously announced Daniel Smith's addition to our Board of Directors, and since then it's been a pleasure to work more closely with him. We are grateful for the time, energy, and perspective he continues to invest in our community.
Applications are currently open for a new role within the Rosetta community - they are looking for a new Science Communications and Engagement Manager. If you are interested, please learn more on our Jobs Board. This is a great opportunity for someone who sits at the intersection of science and community.
We’ve also seen some standout posts on the OMSF blog, alongside a wave of great writing across the broader ecosystem. Here's a look at some of the interesting posts our staff and collaborates made to our blog this year.
- Concerning Uncertainty by Sean Colby
- Deploying Self Hosted Runners on Almost Any Platform by Ethan Holz
- Why We Should Drop (Graph-based) Bond Order and No Longer Use Aromaticity by David Mobley
- A Swiss Army Knife for Conda Issues - Re-create Your Environment by Matt Thompson
- The Free Energy of Everything: Benchmarking OpenFE by Josh Horton
- Dockerize a Tool by Mike Henry
Of course, there are more to come next year! Subscribe to the newsletter here to get these blogs and other announcements straight to your inbox, or follow us on LinkedIn for these updates immediately. If you've already subscribed, Thank You!
Events
In early May 2026, the OMSF community will gather for a series of events in beautiful Barcelona, bringing together our extended family of researchers, developers, and partners from across our ecosystem. These three events take place in close succession, creating a unique opportunity for any one invested in molecular modeling.
Kicking the week off on May 4, the Free Energy Workshop returns for its second year with OMSF, continuing its long-standing role as a focal point for free energy experts worldwide. This year hosted at UPF Campus Ciutadella, registration is now open. If you are interested in sponsoring the event, please contact the organizers at alchemistry@omsf.io
- Dates: May 4th, 5th, and 6th, 2026
- Location: UPF Campus Ciutadella
- Price: $267
- Registration: Submit the your details here.
On May 6th, the team behind OpenFold is hosting the first ever Co-Folding Summit. Focusing on the latest and greatest in co-foldung methods and tools, this summit promises to gather the best minds and models in the field. Not too many details to announce yet, but keep posted here for more registration details.
- Dates: May 6th, 2026
- Location: IEC
- Price: TBA
- Registration: TBA
Finally, closing out the week is our very own OMSF Symposium! This year, we are going back to a two day format, with not just project keynotes, but also workshops and demos scheduled for the second day. It's a perfect way to introduce yourself to our wide range of tools, and to meet the developers behind your favorite packages. We will be opening up registration shortly. If you are interested in learning more, follow us on LinkedIn!
- Dates: May 7th and 8th, 2026
- Location: IEC
- Price: TBA
- Registration: TBA
Beyond this packed week, there are other additional gatherings happening. Our sponsored programs, the Virtual Winter School and Boston Protein Design and Modeling Club, are going full steam ahead in 2026. Follow them to find out more about their meeting schedule.
Project Updates
The past few months have been an exciting time for OMSf software projects. Read on to find out more!
OpenFold
This quarter, the OpenFold Consortium released OpenFold3, a major update to its foundation models for 3D structure prediction. OpenFold3 accurately predicts the three-dimensional structures of proteins, nucleic acids, and small-molecule ligands from sequence alone, enabling more detailed modeling of biomolecular interactions that are central to drug discovery and biomaterials design. Importantly, OpenFold3 is fully open source, allowing researchers and developers to use, adapt, and integrate the model into their own workflows without restriction - a significant step toward democratizing state-of-the-art structure prediction tools.
- Repository: github.com/aqlaboratory/openfold-3
- Documentation: openfold-3.readthedocs.io/en/latest/
Alongside this release, OpenFold has continued to grow its consortium membership, welcoming organizations like Apheris and Johnson & Johnson, who bring expertise in privacy-preserving AI and secure collaborative model training. These partnerships strengthen the consortium’s mission to build open, high-impact AI infrastructure for biological modeling.
OpenADMET
OpenADMET reached a major milestone with the release of its inaugural predictive models for key ADMET properties, including baseline models for CYP3A4 inhibition and PXR activation trained on publicly available ChEMBL data. These models establish an open baseline for ADMET prediction tasks and are now available for use and exploration, advancing transparent, community-driven modeling in drug discovery.
- Repository: github.com/OpenADMET
- Model Workflow and Training: github.com/OpenADMET/openadmet-models
- Tooling Support: github.com/OpenADMET/openadmet-toolkit
In parallel, the OpenADMET team launched the ExpansionRx-OpenADMET Blind Challenge in partnership with Expansion Therapeutics and Hugging Face; a community benchmarking competition where participants predict nine real-world ADMET endpoints from high-quality industry data. With over 100 submissions to date, the challenge has drawn engagement from researchers across academia, industry, and open science communities. Submissions remain open through January 19, 2026, and we’ll share results when the leaderboard closes.
Open Free Energy
Open Free Energy has reached a major milestone with the release of OpenFE v1.7.0, now ready for production use. This update introduces powerful new simulation protocols - including Separated Topologies and Absolute Binding Free Energy workflows - which broaden the range of alchemical free energy calculations the toolkit can perform, all while delivering significant performance improvements over earlier versions.
Alongside the software release, the OpenFE team partnered with industry collaborators to run a large-scale benchmarking study of its relative binding free energy protocols across more than 1,700 ligands. The results show reproducible performance and throughput on par with commercial solutions, demonstrating reproducible performance suitable for real-world drug discovery workflows.
Open Force Field
The Open Force Field team released a prototype version of OpenFF 3.0, Rosemary, representing a significant step toward a unified force field that can model proteins, peptides, and general organic molecules using a single, self-consistent parameter set. While benchmarking and validation are still ongoing, this early release allows users to begin exploring workflows that bridge traditional small-molecule and protein force fields within one framework.
- Benchmarks: zenodo.org/records/17401377
Alongside the force field prototype, the team released an updated post-translational modification (PTM) workflow, showcasing parameterization of modified proteins using the new Pablo tool. Pablo enables flexible chemistry assignment from PDB files, including support for custom and non-standard residues, and integrates directly with the Rosemary force field for end-to-end simulation workflows. Together, these releases highlight Open Force Field’s progress toward more accurate and interoperable modeling of complex biomolecular systems.
- Repository: github.com/openforcefield/ptm_prototype
- Get Started: openff-pablo.readthedocs.io/en/latest/book/01-quickstart.html
Rosetta
The Rosetta Commons has made major strides toward a more accessible software ecosystem this year. Most recently, RFdiffusion3 - the third-generation diffusion-based protein design model - was released as open-source software. RFdiffusion3 represents a major step forward in AI-assisted biomolecular design, enabling the generation of proteins with significantly improved precision and speed compared with earlier versions.
Earlier in the year, RFdiffusion2 was also added to the Rosetta Commons repository, and new supporting tools like AtomWorks and ModelForge continue to expand the ecosystem for structure prediction and modeling.
- Repository: https://github.com/RosettaCommons/foundry
In October, the community gathered for Euro RosettaCon, followed by multiple workshops throughout November. These were complemented by international events including European RosettaCon in Slovenia and ChileCon (AI4PD) in Puerto Varas, as well as Bootcamp++, a hands-on C++ and PyRosetta training held in Nashville. Additional workshops at UC Davis and Vanderbilt rounded out the year.
And the Rosetta Commons have many more coming, with open calls for abstracts already being accepted. Head on over to the Rosetta Commons Events page for the most up-to-date information regarding the communities upcoming events.
Thank you for following and supporting us this year. Whether you are a long time member of our community, or you have just joined us in 2025, we appreciate your time, energy, and attention. We are committed to making 2026 the best year yet. Let's make the most of it!