Form 990 for tax year 2024

Totals

Total revenue

$5,422,298

Total expenses

$5,272,525

Contributions and grants received

$3,661,343

Grants paid

$1,007,681

Program service revenue

$1,758,353

Investment income

$271

Other revenue

$2,331

Total assets, end of year

$3,509,895

Total liabilities, end of year

$2,340,251

Net assets or fund balances, end of year

$1,169,644

Program services expenses (Part IX)

$4,972,344

Management and general expenses

$300,181

Fundraising expenses

$0

Program service expenses (Part III)

$4,972,344

Program services were 94.3% of total expenses calculated — Part IX's program-services column over total expenses, both as filed.

Line items

Every line the filer filed in Parts VIII, IX and X, as filed and in filed order. Blank columns are left out, and the lines are not forced to add up to the totals above.

Mission and programs · 2024

• OMSF welcomed three new projects in 2024 OpenADMET, Alchemistry and Virtual Winter School of Computational Chemistry (VWSCC). Alchemistry and VWSCC are focused on community building efforts in the field of (bio)molecular modeling, more specifically, through organizing events for education, training and networking purposes. OpenADMET is the project with the mission to build open predictive models of safety and toxicity for small molecules to improve humanitys ability to more reliably, cheaply, and effectively treat disease, launched jointly with UCSF and Octant Bio. The OpenADMET project is funded by an ARPA-H grant (the primary awardee is UCSF). The number of staff members and core contributors continued to grow, and we had 27 staff members in 2024. OMSF successfully organized its second annual symposium in Leiden (Netherlands) and successfully gathered more than 100 attendees which included our staff members, industry partners and other members of the molecular modeling community. The list of our hosted projects, their missions and activities is provided below together with our program services and community-oriented efforts. Hosted ProjectsOpenADMETOpenADMET is an open effort to build predictive models of ADMET properties and understand the mechanisms by which they arise. This means systematically characterizing the proteins and mechanisms that give rise to these properties (protein structures and scaled functional and genetic assays), understanding how small molecules interact with these mechanisms (through high-throughput nanoscale chemistry to explore/exploit chemical space), and integrative computational models (AI/ML ligand/structural, mechanistic, and physiological modeling). OMSF is focused on AI/ML ligand and structural modeling, while UCSF and Octant Bios primary role is generation of high-quality datasets. The project was launched on Oct 1, 2024, and this effort is funded by an ARPA-H grant and Gates Foundation. More information about the project and its activities can be found on their website:https://openadmet.org/. OpenFold ConsortiumOpenFold is a non-profit AI research and development consortium developing free and open-source software tools for biology and drug discovery. OpenFolds mission is to bring the most powerful software ever created AI systems with the ability to engineer the molecules of life to everyone. These tools can be used by academics, biotech and pharmaceutical companies, or students learning to create the medicines of tomorrow, to accelerate basic biological research, and bring new cures to market that would be impossible to discover without AI. This project aims to:Collaboratively develop a permissively licensed model competitive with the performance of state-of-the-art models. Provide the entire training & inference stack and training datasets under the same permissive license Optimize the performance of this model for use on state-of-the-art and widely available GPUs. In 2024, the OpenFold Consortium started working on reproducing the AlphaFold3 model and releasing the resulting code, weights and data under permissive licenses. The consortium has grown from 10 to 14 industry members between 2023 and 2024. The list of current OpenFold members, links to software repositories and latest news about their activities can be found on their website (https://openfold.io). Open Force Field ConsortiumThe Open Force Field Consortium (OpenFF) is a network of academic and industry researchers working together to advance science and infrastructure required for building the next generation of small molecule and biomolecular force fields. The shared goal of these researchers is to develop automated and systematic data-driven techniques to parameterize and assess new generations of more accurate force fields. Software and data are released under open licensing agreements to enable rapid application, validation, extension and any kind of modification by our users and contributors. In addition, the Initiative aims to build and support a strong community of users and contributors from industry and academia. The OpenFF Sage 2.2 release delivered targeted improvements to critical pharmaceutical chemistries. Independent validation demonstrated that OpenFF outperforms leading commercial packages on challenging BACE1 scaffold-hopping transformations, confirming the force field's competitive accuracy. A major new feature is the preview release of a graph neural network model for ultra-fast, conformation-independent charge assignment that operates 10-100x faster than traditional methods with improved accuracy and elimination of conformer-dependent errors. This method scales seamlessly to large biomolecules including proteins and polymers, enabling self-consistent parameterization across all molecular types including non-natural amino acids. Infrastructure development released the production-ready Alchemiscale platform completing over 100,000 binding free energy calculations on free academic compute resources and being adopted extensively by OpenFF, OpenFE, and the ASAP Antiviral Drug Discovery Consortium for routine force field testing. The YAMMBS automated benchmarking system enabled testing of 75+ different force field fits over the past year while identifying and correcting systematic errors through continuous validation of molecular geometries and energetics. Significant progress was made toward the self-consistent protein + small molecule force field ("Rosemary"), with the team generating requisite peptide quantum chemical data and achieving performance matching AMBER ff14SB on specific benchmarks; a recent breakthrough in fitting strategy using trajectory reweighting has dramatically accelerated progress, putting the release on track for the upcoming project year to enable truly accurate protein-ligand binding with fully consistent parameterization. Additional technical developments included completion of virtual site infrastructure, advancement of automated chemical perception using BESMARTS for data-driven typing decisions, development of a direct polarizability model (AM1-BCC-dPol), biopolymer loading infrastructure for diverse polymers and materials applications, and maintenance of stable toolkit API enabling consistent workflow integration. The team also significantly enhanced documentation with centralized examples and live demonstration sessions that showcase the ease and versatility of OpenFF tools. Open Force Field had 17 industry members in 2024. You can find more information about the Consortiums activities, members, resulting publications and software repositories on their website: https://openforcefield.org.Open Free Energy ConsortiumOpen Free Energy Consortium (OpenFE) provides essential tools for drug discovery and design, supporting free energy calculations that are robust and scalable without commercial license fees. OpenFE bridges a gap in the landscape of free energy simulation software, developing stable, reliable, and openly licensed software. The project's flagship accomplishment was the launch of openfe v1.0 in April 2024, marking the first stable release with guaranteed maintenance of functionality and behavior. Validation efforts demonstrated promising accuracy, with partner testing showing comparable performance to industry-standard proprietary methods across 12 transformations, while a comprehensive benchmarking study successfully evaluated 211 ligands across 9 diverse protein systems. Major technical developments expanded the toolkit's capabilities to include charge-changing transformations between ligands with different net charges, core atom element changes, spectator molecule support for cofactors, and automated post-simulation analysis for identifying problematic calculations through energetics and structural assessments. Significant usability improvements included comprehensive documentation redesign, enhanced CLI tools for streamlined network planning and results gathering, tutorial development, and a remarkable 10-60x performance improvement in the LOMAP atom mapping tool. A critical achievement involved identifying and addressing a major AmberTools partial charge assignment bug that was causing large variability in free energy predictionsthe team traced this issue to inconsistent conformer selection across different computing environments and unstable partial charge assignments, then worked collaboratively with upstream developers to implement fixes that benefit the entire computational chemistry community. In 2023, Open Free Energy had 17 industry partners. The list of OpenFE products, industry partners, and software repositories can be found on their website: https://openfree.energy. Open RosettaOpen Rosetta is the project exploring the transition pathways for the Rosetta Commons to an open source model under the OMSF umbrella, with the initial focus on the community building efforts and new developments. The Rosetta Commons are a community of over 100 labs worldwide who coll

  Expenses $4,972,344 · Grants $1,007,681

As filed on the 2024 return.

Listed in Part VII

• Marcus Wieder

  ML Scientist

  Total calculated

  $208,101

  From this org

  $208,101

  From related orgs

  $0

  Other

  $0

  Hours/week

  40

  Schedule J breakdown matched by us

  Base

  $208,101

  Bonus

  —not reported

  Other

  —not reported

  Deferred

  —not reported

  Nontaxable benefits

  —not reported

  Filer's total

  $208,101

• Julia Koehler

  Project Director

  Total calculated

  $186,678

  From this org

  $176,313

  From related orgs

  $0

  Other

  $10,365

  Hours/week

  40

  Schedule J breakdown matched by us

  Base

  $176,313

  Bonus

  —not reported

  Other

  —not reported

  Deferred

  —not reported

  Nontaxable benefits

  $10,365

  Filer's total

  $186,678

• Jennifer Wei

  Software Scientist

  Total calculated

  $148,717

  From this org

  $140,000

  From related orgs

  $0

  Other

  $8,717

  Hours/week

  40

• Irfan Alibay

  Software Scientist

  Total calculated

  $145,061

  From this org

  $145,061

  From related orgs

  $0

  Other

  $0

  Hours/week

  40

• David Swenson

  Software Scientist

  Total calculated

  $143,211

  From this org

  $132,875

  From related orgs

  $0

  Other

  $10,336

  Hours/week

  40

• Karmen Condic-Jurkic

  Sec/Treas/ED

  Total calculated

  $102,098

  From this org

  $46,667

  From related orgs

  $0

  Other

  $55,431

  Hours/week

  40

• David Mobley

  President

  Total calculated

  $0

  Hours/week

  2

• Brian Weitzner

  Director

  Total calculated

  $0

  Hours/week

  0.5

• Clara Christ

  Director

  Total calculated

  $0

  Hours/week

  0.2

• Jeffrey Gray

  Director

  Total calculated

  $0

  Hours/week

  0.2

Compensation is as filed in Part VII: reportable pay from this organization and from related organizations, other compensation, and hours per week. The total, marked calculated, is our sum of the three dollar columns, with a blank column counted as $0; the form itself has no total. On a Form 990-EZ the columns are Part IV's instead, and on a Form 990-PF Part VIII's — compensation, contributions to benefit plans, and expense account — and the total sums those. A Schedule J breakdown is as filed: base pay, bonus, other, deferred compensation and nontaxable benefits from this organization, with any from related organizations beside each, and the filer's own total. It is shown with the Part VII listing of the same name on the same return. The form gives no key between the two, so that pairing is ours, and is marked matched by us.

Schedule J

Part I's answers as filed. A line not listed was left unanswered; on lines 1a and 3 that means the box was not checked.

Grants on this return

• Rosetta Commons Foundation

  $863,000

  Software Platform Development

• UC Irvine

  $35,000

  Research

• Johns Hopkins University

  $32,522

  Research

• Arizona State University

  $7,069

  Open Source Ecosystem Development

Provenance

Return type

form_990

Tax year

2024

Filed

17 November 2025

IRS schema version

2024v5.2

Source object

202513219349319151

Check it against the return as filed: the GivingTuesday Data Lake hosts a public copy of the original XML.

The source document this was parsed from is served over the API, which is where bulk and document access live.