David Baker’s lab at the University of Washington is announcing two major leaps in the field of AI-powered protein design. The first is a souped-up version of its existing RFdiffusion2 tool that can now design enzymes with performance nearly on par with those found in nature. The second is the release of a new, general-purpose version of its model, named RFdiffusion3, which the researchers are calling their most powerful and versatile protein engineering technology to date.

Last year, Baker received the Nobel Prize in Chemistry for his pioneering work in protein science, which includes a deep-learning model called RFdiffusion. The tool allows scientists to design novel proteins that have never existed. These machine-made proteins hold immense promise, from developing medicines for previously untreatable diseases to solving knotty environmental challenges.

Baker leads the UW’s Institute for Protein Design, which released the first version of the core technology in 2023, followed by RFdiffusion2 earlier this year. The second model was fine-tuned for creating enzymes — proteins that orchestrate the transformation of molecules and dramatically speed up chemical reactions.

The latest accomplishments are being shared today in publications in the leading scientific journals Nature and Nature Methods, as well as a preprint last month on bioRxiv.

A better model for enzyme construction

In the improved version of RFdiffusion2, the researchers took a more hands-off approach to guiding the technology, giving it a specific enzymatic task to perform but not specifying other features. Or as the team described it in a press release, the tool produces “blueprints for physical nanomachines that must obey the laws of chemistry and physics to function.”

“You basically let the model have all this space to explore and … you really allow it to search a really wide space and come up with great, great solutions,” said Seth Woodbury, a graduate student in Baker’s lab and author on both papers publishing today.

In addition to UW scientists, researchers from MIT and Switzerland’s ETH Zurich contributed to the work.

The new approach is remarkable for quickly generating higher-performing enzymes. In a test of the tool, it was able to solve 41 out of 41 difficult enzyme design challenges, compared to only 16 for the previous version.

“When we designed enzymes, they’re always an order of magnitude worse than native enzymes that evolution has taken billions of years to find,” said Rohith Krishna, a postdoctoral fellow and lead developer of RFdiffusion2. “This is one of the first times that we’re not one of the best enzymes ever, but we’re in the ballpark of native enzymes.”

The researchers successfully used the model to create proteins calls metallohydrolases, which accelerate difficult reactions using a precisely positioned metal ion and an activated water molecule. The engineered enzymes could have important applications, including the destruction of pollutants.

The promise of rapidly designed catalytic enzymes could unleash wide-ranging applications, Baker said.

“The first problem we really tackled with AI, it was largely therapeutics, making binders to drug targets,” he said. “But now with catalysis, it really opens up sustainability.”

The researchers are also working with the Gates Foundation to figure out lower-cost ways to build what are known as small molecule drugs, which interact with proteins and enzymes inside cells, often by blocking or enhancing their function to effect biological processes.

The most powerful model to date

While RFdiffusion2 is fine-tuned to make enzymes, the Institute for Protein Design researchers were also eager to build a tool with wide-ranging functionality. RFdiffusion3 is that new AI model. It can create proteins that interact with virtually every type of molecule found in cells, including the ability to bind DNA, other proteins and small molecules, in addition to enzyme-related functions.

“We really are excited about building more and more complex systems, so we didn’t want to have bespoke models for each application. We wanted to be able to combine everything into one foundational model,” said Krishna, a lead developer of RFdiffusion3.

Today the team is publicly releasing the code for the new machine learning tool.

“We’re really excited to see what everyone else builds on it,” Krishna said.

And while the steady stream of model upgrades, breakthroughs and publications in top-notch journals seems to continue unabated from the Institute for Protein Design, there are plenty of behind-the-scenes stumbles, Baker said.

“It all sounds beautiful and simple at the end when it’s done,” he said. “But along the way, there’s always the moments when it seems like it won’t work.”

But the researchers keep at it, and so far at least, they keep finding a path forward. And the institute continues minting new graduates and further training postdocs who go on to launch companies or establish their own academic labs.

“I don’t surf, but I sort of feel like we’re riding a wave and it’s just fun,” Baker said. “I mean, it’s so many, so many problems are getting solved. And yeah, it’s really exhilarating, honestly.”

The Nature paper, titled “Computational design of metallohydrolases,” was authored by Donghyo Kim, Seth Woodbury, Woody Ahern, Doug Tischer, Alex Kang, Emily Joyce, Asim Bera, Nikita Hanikel, Saman Salike, Rohith Krishna, Jason Yim, Samuel Pellock, Anna Lauko, Indrek Kalvet, Donald Hilvert and David Baker.

The Nature Methods paper, titled “Atom-level enzyme active site scaffolding using RFdiffusion2,” was authored by Woody Ahern, Jason Yim, Doug Tischer, Saman Salike, Seth Woodbury, Donghyo Kim, Indrek Kalvet, Yakov Kipnis, Brian Coventry, Han Raut Altae-Tran, Magnus Bauer, Regina Barzilay, Tommi Jaakkola, Rohith Krishna and David Baker.

A Seattle biotech startup born from a Nobel laureate’s lab has landed $12.7 million and partnerships with pharmaceutical giants Pfizer and Kite Pharma by using AI to design proteins that mount a multi-pronged attack on diseases.

Accipiter Biosciences emerged from stealth today with a leadership team that includes researchers who worked at the University of Washington’s Institute for Protein Design under David Baker, a 2024 Nobel Prize in Chemistry winner for his breakthroughs in building proteins from scratch.

The company is using artificial intelligence tools developed at the institute to engineer de novo proteins that have the unusual ability to bind multiple cellular targets at once, potentially amplifying their illness-fighting impact.

“We want to establish this new modality,” said Matthew Bick, Accipiter Bio’s co-founder and CEO. The strategy, he added, could unlock new ways to more effectively treat complicated diseases.

There’s evidence that combinations of drugs sometimes perform better than single therapies, but the challenge has been coordinating their actions so they work together at the same location.

In some forms of cancer, for example, multiple cell functions need to be turned on simultaneously to produce helpful molecules that work synergistically to create an effect “that is not just additive, it’s multiplicative,” Bick said.

The approach could also speed U.S. Food and Drug Administration approval and cut costs. Typically, when two drugs are combined to treat a condition, each must undergo its own expensive Phase 1 safety trial, followed by an additional trial testing them together. A single multi-functional drug would need just one Phase 1 trial.

Multiple avenues to drug therapies

Accipiter Bio has entered into a collaboration and license agreement with Pfizer to research and engineer new molecules. The deal provides an upfront payment for the startup and the potential to earn more than $330 million if Accipiter Bio hits certain milestones and through royalties.

“With Accipiter’s platform technology and collaboration, Pfizer aims to solve complex therapeutic problems with biologics that may have previously been unattainable,” said Jeffrey Settleman, Pfizer Oncology R&D’s chief scientific officer.

Accipiter Bio also has an agreement with the oncology drug company Kite, which is owned by Gilead Sciences, to design proteins for use in cell therapies. The arrangement similarly includes initial funding with the possibility of milestone payments and royalties. Kite has the option of acquiring molecules created through the arrangement and develop them into therapeutics for global sales.

On top of those efforts, Accipiter Bio has four of its own drug-development programs. Two programs are preparing for formal FDA discussions about human testing — a stage called pre-IND .

Bick would not provide details on the efforts, but said the company is researching agents for treating cancers and irritable bowel syndrome, among other ailments.

Funding and leadership

Flying Fish Partners and Takeda co-led the seed round. Additional investors are Columbus Venture Partners, Cercano Capital, Washington Research Foundation, Alexandria Investments, Pack Ventures and Argonautic Ventures.

“We’ve reached the point where computation isn’t just speeding up biology,” said Heather Gorham, principal at Flying Fish Partners and Accipiter board member. “It’s expanding what’s biologically possible.”

The startup launched in March 2023 and previously raised about $800,000 to get off the ground. Bick was a senior fellow in Baker’s lab for more than seven years and later a senior director for Seattle’s Neoleukin Therapeutics.

Accipiter Bio has 17 employees. The leadership team has three members in addition to Bick.

• Javier Castellanos, co-founder and chief technologist, was a graduate student with Baker; co-founder and CTO of Cyrus Biotechnology, another protein design startup; and a past director at Neoleukin.
• Hector Rincon-Arano, co-founder and chief scientist, was with Seagen (now a division of Pfizer) for more than seven years where he helped take a therapeutic from proof-of-concept to the first step of getting a new drug approved. He was also briefly at Neoleukin.
• William Canestaro, chief operating officer and chief strategy officer, has worked on the business and investing side of biotech with roles at the UW’s Michael G. Foster School of Business, Washington Research Foundation, Pack Ventures, Pioneer Square Labs, Cyclera Therapeutics and others. He has served on the board of directors for multiple startups.

Building on experience

While the strategy of using AI to build a new class of proteins could open the door to groundbreaking therapies, drug development is a risky business.

Neoleukin was a biotech company co-founded by Baker that spun out of the UW in 2019. The startup’s lead drug candidate, an engineered protein used in cancer treatment, under-performed in a Phase 1 trial. Neoleukin laid off many of its employees before merging with another company.

The three co-founders met at the startup and gained valuable technical and strategic lessons from the experience, Bick said. That included the need to have multiple drug programs running at once and insights into preventing immunogencity, which is an unwanted immune response to foreign bodies.

“We were part of the team,” he said, “that took the first fully de novo protein into patients.”

Researchers from Nobel Laureate David Baker’s lab and the University of Washington’s Institute for Protein Design (IPD) have used artificial intelligence to design antibodies from scratch — notching another game-changing breakthrough for the scientists and their field of research.

“It was really a grand challenge — a pipe dream,” said Andrew Borst, head of electron microscopy R&D at IPD. Now that they’ve hit the milestone of engineering antibodies that successfully bind to their targets, the research “can go on and it can grow to heights that you can’t imagine right now.”

Borst and his colleagues are publishing their work in the peer-reviewed journal Nature. The development could supercharge the $200 billion antibody drug industry.

Before the advent of AI-based tools, scientists made antibodies by immunizing animals and hoping they would produce useful molecules. The process was laborious and expensive, but tremendously important. Many powerful new drugs for treating cancer and autoimmune diseases are antibody-based, using the proteins to hit specific targets.

Baker, who won the Nobel Prize in Chemistry last year, was recognized for his work unraveling the molecular design of proteins and developing AI-powered tools to rapidly build and test new ones. The technology learns from existing proteins and how they function, then creates designs to solve specific challenges.

In the new research, the team focused on the six loops of protein on the antibody’s arms that serves as fingers that grab its target. Earlier efforts would tweak maybe one of the loops, but the latest technology allows for a much bigger play.

“We are starting totally from scratch — from the loop perspective — so we’re designing all six,” said Robert Ragotte, a postdoctoral researcher at IPD. “But the rest of the antibody, what’s called the framework, that is actually staying the same.”

The hope is that by retaining the familiar humanness of most of the antibody, a patient’s immune system would ignore the drug rather than mount an offense against an otherwise foreign molecule.

The researchers tested their computer creations against multiple real-world targets including hemagglutinin, a protein on flu viruses that allow them to infect host cells; a potent toxin produced by the C. difficile bacteria; and others.

The lab tests showed that in most cases, the new antibodies bound to their targets as the online simulations predicted they would.

“They were binding in the right way with the right shape against the right target at the spot of interest that would potentially be useful for some sort of therapeutic effect,” Borst said. “This was a really incredible result to see.”

Borst added that the computational and wet lab biologists worked closely together, allowing the scientists to refine their digital designs based on what the real-life experiments revealed.

The software used to create the antibodies is freely available on GitHub for anyone to use. Xaira Therapeutics, a well-funded biotech startup led by IPD alumni, has licensed some of the technology for its commercial operations and multiple authors on the Nature paper are currently employed by the company.

While the antibodies created as part of the research demonstrated the software’s potential, there are many more steps to engineering a potential therapy. Candidate drugs need to be optimized for additional features such as high solubility, a strong affinity for a target and minimizing immunogenicity — which is an unwanted immune response.

Before joining IPD four years ago, Ragotte was a graduate student doing conventional antibody discovery and characterization using animals.

The idea that one day you could get on a computer, choose a target, and create a DNA blueprint for building a protein was almost unimaginable, he said. “We would talk about it, but it didn’t even seem like a tractable problem at that point.”

The Nature study is titled “Atomically accurate de novo design of antibodies with RFdiffusion.” The lead authors include Nathaniel Bennett, Joseph Watson, Robert Ragotte, Andrew Borst, DéJenaé See,
Connor Weidle and Riti Biswas, all of whom were affiliated with the UW at the time the research was conducted, and Yutong Yu of the University of California, Irvine. David Baker is the senior author.

Additional authors are: Ellen Shrock, Russell Ault, Philip Leung, Buwei Huang, Inna Goreshnik, John Tam, Kenneth Carr, Benedikt Singer, Cameron Criswell, Basile Wicky, Dionne Vafeados, Mariana Sanchez, Ho Kim, Susana Torres, Sidney Chan, Shirley Sun, Timothy Spear, Yi Sun, Keelan O’Reilly, John Maris, Nikolaos Sgourakis, Roman Melnyk and Chang Liu.

Seattle and the Nordic nations have strong ties — from Norway’s first astronaut launching on a SpaceX mission to Scandinavian companies supporting the efforts to electrify the region’s ferries. And those connections stretch back more than a century when John Nordstrom, a Swedish immigrant, co-founded an eponymously named shoe store in Seattle’s downtown.

“As we look towards the future, we build on our history and also our heritage. But today, we reach for much more. We are partners in innovation, from high tech in the AI revolution to building more security projects,” said Anniken Huitfeldt, Norway’s ambassador to the United States.

On Tuesday evening, Huitfeldt helped kick off the annual Nordic Innovation Summit, held in Seattle’s Ballard neighborhood at the National Nordic Museum.

The event continues through Wednesday and features speakers and panelists discussing clean energy topics such as data center operations, microgrids and recycling; cyber security initiatives; innovation in biotech; and sustainable transportation.

“We’ve seen innovations come out of the summit,” said Seattle City Councilmember Dan Strauss, speaking Tuesday. “The ability to integrate technology into new business practices is something that has happened here before, and I’m excited to see it occur here again.”

Other speakers at the opening session included University of Washington Nobel Laureate David Baker, who shared new research published since he received his award from the Swedish institution one year ago. Baker is leading efforts that harness AI to create proteins that can be used in health care, environmental cleanup and other challenges.

Norwegian Astronaut Jannicke Mikkelsen livestreamed a presentation from Oslo to share the story of her journey to space, which began at age 10 when she had a devastating horse riding accident. During a five-year recovery, Mikkelsen became obsessed with NASA — even calling the agency’s Johnson Space Center seeking employment as a 12-year-old.

“It cost us a fortune,” she said. “My dad can’t afford to drive his car to work for a week because we can’t afford gas because I called the U.S. and I applied for a job.”

Mikkelsen ultimately pursued a career in 3D photography and 3D filmmaking, residing in Svalbard, a frozen tundra that ranks as one of the world’s northernmost inhabited areas.

“This is a place that equipped me the best to become an astronaut, because it’s sort of fear-based learning,” she said. “As soon as you leave town, any mistake you make could potentially kill you. It’s exceptionally cold. There is no communication to the outside world as you leave town.”

In March of this year, SpaceX — which has significant manufacturing operations in the Seattle region — launched Mikkelsen and three other non-governmental astronauts on the first crewed orbit over the Earth’s polar regions. The Fram2 mission spanned 3½ days. The astronauts performed scientific research, including data collection on the Northern Lights, which are beautiful but poorly understood and can destroy satellites.

Mikkelsen highlighted the importance of the collaboration between nations.

“We have four non-American astronauts on board an American spaceship. This, to me, just shows great trust that the U.S. showed in us,” she said. “But it’s also American technology that flew the first Norwegian astronaut into space.”

Speakers participating in the summit on Wednesday include Mathias Sundin, founder of the Warp Institute Foundation; Douglas Kieper, director of the Paul G. Allen Research Center; Nick Huzar, co-founder of OfferUp; Sunil Gottumukkala, CEO of Averlon; Petri Hautakangas, CEO of Tupl; Maiken Møller-Hansen, director of energy and sustainability for Amazon Devices; and other corporate, government and academic leaders.