MONROVIA, Calif. — The laboratory at Terray Therapeutics is a symphony of miniaturized automation. The robots spin quickly and carry small tubes of liquid to their stations. Scientists in blue coats, sterile gloves and protective glasses monitor the machines.
But what really happens is at the nanoscale: Proteins in solution combine with chemical molecules held in tiny wells in special silicon chips that resemble microscopic muffin tins. Millions and millions of every interaction is recorded every day, producing 50 terabytes of raw data per day; That’s the equivalent of more than 12,000 movies.
The laboratory, about two-thirds the size of a football field, is a data factory for AI-powered drug discovery and development in Monrovia, California. This is part of a wave of young companies and startups trying to use artificial intelligence to produce more effective drugs faster.
Sign up for The Morning newsletter from The New York Times
To reimagine drug discovery, companies are leveraging new technology that learns from large amounts of data to generate answers. They move the field from painstaking craftsmanship to more automated precision; A change supported by artificial intelligence that learns and becomes smarter.
“When you have the right kind of data, AI can work and get really good results,” said Jacob Berlin, co-founder and CEO of Terray.
Many of the first commercial uses of generative AI, which can produce everything from poetry to computer programs, were to help eliminate the drudgery of routine office tasks, customer service, and coding. But drug discovery and development is a huge industry that experts say is ripe for AI transformation.
Artificial intelligence is a “once-in-a-century opportunity” for the pharmaceutical industry, according to consulting firm McKinsey & Co.
Just as popular chatbots like ChatGPT are trained on text from the internet, and image generators like DALL-E learn from large amounts of images and videos, AI for drug discovery relies on data. And these are very specific data; measurements of molecular information, protein structures and biochemical interactions. AI suggests potentially useful drug candidates by exploiting patterns in the data, as if it were matching chemical keys with the right protein locks.
Because AI for drug development is backed by rigorous scientific data, the likelihood of toxic “hallucinations” is much less than with more broadly trained chatbots. And any potential drug must undergo extensive testing in laboratories and clinical trials before it is approved for patients.
Companies like Terray are building large high-tech laboratories to produce information that will help train AI; This allows for rapid experimentation and the ability to identify patterns and make predictions about what might work.
Generative AI can then digitally design a drug molecule. This design is converted into a physical molecule in a high-speed automated laboratory and tested for its interaction with the target protein. Positive or negative results are recorded and fed back to the AI software to improve the next design, thus speeding up the overall process.
Although some drugs developed by artificial intelligence are in clinical trials, it is still early days.
“Generative AI is transforming the field, but the drug development process is complex and very human,” said David Baker, a biochemist and director of the Institute for Protein Design at the University of Washington.
Drug development has traditionally been an expensive, time-consuming, haphazard endeavor. Studies on the cost of designing a drug and guiding clinical trials through to final approval vary widely. However, the total cost is estimated to be around 1 billion dollars. It takes 10-15 years. Almost 90% of candidate drugs that enter human clinical trials fail, often due to lack of effectiveness or unforeseen side effects.
Young AI drug developers are trying to use their technology to improve these possibilities while saving time and money.
The most consistent sources of funding come from pharmaceutical giants, which have long served as partners and bankers of smaller research ventures. Today’s AI drugmakers are often focused on accelerating preclinical development stages, which traditionally take 4-7 years. Some may try to enter clinical trials themselves. But this is the stage where big pharmaceutical companies often take over and conduct expensive human trials that can take another seven years.
For established pharmaceutical companies, the partnership strategy is a relatively low-cost way to leverage innovation.
“For them, it’s like taking an Uber to take you somewhere instead of buying a car,” said Gerardo Ubaghs Carrión, a former biotechnology investment banker at Bank of America Securities.
Big pharmaceutical companies pay research partners to reach milestones on the road to drug candidates; This figure can reach hundreds of millions of dollars over the years. And if a drug is eventually approved and becomes a commercial success, a royalty revenue stream emerges.
Companies such as Terray, Recursion Pharmaceuticals, Schrödinger and Isomorphic Labs are pursuing breakthroughs. However, in general, there are two different ways; those that are intended to build large laboratories and those that are not.
Isomorphic, the drug discovery product of the technology giant’s central artificial intelligence group Google DeepMind, takes the view that the better the artificial intelligence, the less data will be needed. And he is confident in his software skills.
In 2021, Google DeepMind released software that accurately predicts the shapes in which amino acid sequences will fold into proteins. These 3-dimensional shapes determine how a protein works. Because proteins drive the behavior of all living things, this has strengthened biological understanding and aided drug discovery.
Last month, Google DeepMind and Isomorphic announced that their newest AI model, AlphaFold 3, can predict how molecules and proteins will interact; This is the next step in drug design.
“We focus on the computational approach,” said Max Jaderberg, Isomorphic’s Director of Artificial Intelligence. “We think there is huge potential that can be unlocked.”
Terray, like most drug development initiatives, is a byproduct of years of scientific research combined with newer advances in artificial intelligence.
Berlin, who received his doctorate in chemistry from Caltech, has followed advances in nanotechnology and chemistry throughout his career. Terray grew out of an academic project started more than a decade ago at the City of Hope cancer center near Los Angeles, where Berlin has a research group.
Terray is focusing on developing small molecule drugs, that is, any drug that a person can take in pills, such as aspirin and statins. The pills are easy to take and cheap to produce.
Terray’s fancy labs are a far cry from the old days in academia, when data was stored in Excel spreadsheets and automation was a distant goal.
“I was a robot,” recalled Kathleen Elison, Terray co-founder and senior scientist.
But when Terray was founded in 2018, the technologies needed to create the industrial-style data lab were advancing rapidly. Terray relied on advances from outside manufacturers to make the microscale chips Terray designed. Their labs are full of automated equipment, but almost all of them are customized thanks to advances in 3D printing technology.
The Terray team was aware from the beginning that AI would be vital in making sense of data stores, but it was only later that the potential for generative AI in drug development became apparent – though not before ChatGPT became a big hit in 2022.
Narbe Mardirossian, a senior scientist at Amgen, became Terray’s chief technology officer in 2020, in part because of the wealth of data produced in the lab. Under Mardirossian, Terray built data science and AI teams and built an AI model to transform chemical data into mathematics and back again. The company has released an open source version.
Terray has partnership agreements with Bristol Myers Squibb and Calico Life Sciences, a subsidiary of Google parent company Alphabet that focuses on age-related diseases. The terms of these agreements were not disclosed.
To expand, Terray will need funds beyond its $80 million in venture financing, said Eli Berlin, Jacob Berlin’s younger brother. He said he left his job in private equity to become the startup’s co-founder and chief financial officer, convinced that technology could open the door to a lucrative business.
Terray is developing new drugs for inflammatory diseases such as lupus, psoriasis and rheumatoid arthritis. Jacob Berlin said the company expects the drugs to be submitted to clinical trials in early 2026.
It may speed up Terray and his colleagues’ drug-making innovations, but that’s about it.
“The ultimate test for us and for the field in general is whether you can look back in 10 years and say that the clinical success rate has increased a lot and we have better drugs for human health,” Berlin said.
c.2024 New York Times Corporation