December 30, 2024

How A.I. Is Revolutionizing Drug Development

The laboratory at Terray Therapeutics is a symphony of miniaturized automation. Robots whir, shuttling tiny tubes of fluids to their stations. Scientists in blue coats, sterile gloves and protective glasses monitor the machines.

But the real action is happening at nanoscale: Proteins in solution combine with chemical molecules held in minuscule wells in custom silicon chips that are like microscopic muffin tins. Every interaction is recorded, millions and millions each day, generating 50 terabytes of raw data daily — the equivalent of more than 12,000 movies.

The lab, about two-thirds the size of a football field, is a data factory for artificial-intelligence-assisted drug discovery and development in Monrovia, Calif. It’s part of a wave of young companies and start-ups trying to harness A.I. to produce more effective drugs, faster.

The companies are leveraging the new technology — which learns from huge amounts of data to generate answers — to try to remake drug discovery. They are moving the field from a painstaking artisanal craft to more automated precision, a shift fueled by A.I. that learns and gets smarter.

“Once you have the right kind of data, the A.I. can work and get really, really good,” said Jacob Berlin, co-founder and chief executive of Terray.

Most of the early business uses of generative A.I., which can produce everything from poetry to computer programs, have been to help take the drudgery out of routine office tasks, customer service and code writing. Yet drug discovery and development is a huge industry that experts say is ripe for an A.I. makeover.

A.I. is a “once-in-a-century opportunity” for the pharmaceutical business, according to the consulting firm McKinsey & Company.

Just as popular chatbots like ChatGPT are trained on text across the internet, and image generators like DALL-E learn from vast troves of pictures and videos, A.I. for drug discovery relies on data. And it is very specialized data — molecular information, protein structures and measurements of biochemical interactions. The A.I. learns from patterns in the data to suggest possible useful drug candidates, as if matching chemical keys to the right protein locks.

Because A.I. for drug development is powered by precise scientific data, toxic “hallucinations” are far less likely than with more broadly trained chatbots. And any potential drug must undergo extensive testing in labs and in clinical trials before it is approved for patients.

Companies like Terray are building big high-tech labs to generate the information to help train the A.I., which enables rapid experimentation and the ability to identify patterns and make predictions about what might work.

Generative A.I. can then digitally design a drug molecule. That design is translated, in a high-speed automated lab, to a physical molecule and tested for its interaction with a target protein. The results — positive or negative — are recorded and fed back into the A.I. software to improve its next design, accelerating the overall process.

While some A.I.-developed drugs are in clinical trials, it’s still early days.

“Generative A.I. is transforming the field, but the drug-development process is messy 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, hit-or-miss endeavor. Studies of the cost of designing a drug and navigating clinical trials to final approval vary widely. But the total expense is estimated at $1 billion on average. It takes 10 to 15 years. And nearly 90 percent of the candidate drugs that enter human clinical trials fail, usually for lack of efficacy or unforeseen side effects.

The young A.I. drug developers are striving to use their technology to improve those odds, while cutting time and money.

Their most consistent source of funding comes from the pharma giants, which have long served as partners and bankers to smaller research ventures. Today’s A.I. drugmakers are typically focused on accelerating the preclinical stages of development, which have conventionally taken four to seven years. Some may try to go into clinical trials themselves. But that stage is where major pharma corporations usually take over, operating the expensive human trials, which can take another seven years.

For the established drug companies, the partner strategy is a relatively low-cost path to tap innovation.

“For them, it’s like taking an Uber to get you somewhere instead of having to buy a car,” said Gerardo Ubaghs Carrión, a former biotech investment banker at Bank of America Securities.

The major pharma companies pay their research partners for reaching milestones toward drug candidates, which can reach hundreds of millions of dollars over years. And if a drug is eventually approved and becomes a commercial success, there is a stream of royalty income.

Companies like Terray, Recursion Pharmaceuticals, Schrödinger and Isomorphic Labs are pursuing breakthroughs. But there are, broadly, two different paths — those that are building big labs and those that aren’t.

Isomorphic, the drug discovery spinout from Google DeepMind, the tech giant’s central A.I. group, takes the view that the better the A.I., the less data that’s needed. And it is betting on its software prowess.

In 2021, Google DeepMind released software that accurately predicted the shapes that strings of amino acids would fold into as proteins. Those three-dimensional shapes determine how a protein functions. That was a boost to biological understanding and helpful in drug discovery, since proteins drive the behavior of all living things.

Last month, Google DeepMind and Isomorphic announced that their latest A.I. model, AlphaFold 3, can predict how molecules and proteins will interact — a further step in drug design.

“We’re focusing on the computational approach,” said Max Jaderberg, chief A.I. officer at Isomorphic. “We think there is a huge amount of potential to be unlocked.”

Terray, like most of the drug development start-ups, is a byproduct of years of scientific research combined with more recent developments in A.I.

Dr. Berlin, the chief executive, who earned his Ph.D. in chemistry from Caltech, has pursued advances in nanotechnology and chemistry throughout his career. Terray grew out of an academic project begun more than a decade ago at the City of Hope cancer center near Los Angeles, where Dr. Berlin had a research group.

Terray is concentrating on developing small-molecule drugs, essentially any drug a person can ingest in a pill like aspirin and statins. Pills are convenient to take and inexpensive to produce.

Terray’s sleek labs are a far cry from the old days in academia when data was stored on Excel spreadsheets and automation was a distant aim.

“I was the robot,” recalled Kathleen Elison, a co-founder and senior scientist at Terray.

But by 2018, when Terray was founded, the technologies needed to build its industrial-style data lab were progressing apace. Terray has relied on advances by outside manufacturers to make the micro-scale chips that Terray designs. Its labs are filled with automated gear, but nearly all of it is customized — enabled by gains in 3-D printing technology.

From the outset, the Terray team recognized that A.I. was going to be crucial to make sense of its stores of data, but the potential for generative A.I. in drug development became apparent only later — though before ChatGPT became a breakout hit in 2022.

Narbe Mardirossian, a senior scientist at Amgen, became Terray’s chief technology officer in 2020 — in part because of its wealth of lab-generated data. Under Dr. Mardirossian, Terray has built up its data science and A.I. teams and created an A.I. model for translating chemical data to math, and back again. The company has released an open-source version.

Terray has partnership deals with Bristol Myers Squibb and Calico Life Sciences, a subsidiary of Alphabet, Google’s parent company, that focuses on age-related diseases. The terms of those deals are not disclosed.

To expand, Terray will need funds beyond its $80 million in venture funding, said Eli Berlin, Dr. Berlin’s younger brother. He left a job in private equity to become a co-founder and the start-up’s chief financial and operating officer, persuaded that the technology could open the door to a lucrative business, he said.

Terray is developing new drugs for inflammatory diseases including lupus, psoriasis and rheumatoid arthritis. The company, Dr. Berlin said, expects to have drugs in clinical trials by early 2026.

The drugmaking innovations of Terray and its peers can speed things up, but only so much.

“The ultimate test for us, and the field in general, is if in 10 years you look back and can say the clinical success rate went way up and we have better drugs for human health,” Dr. Berlin said.