Entrenched in academia, chemist Jacob Berlin spent a decade making small molecules to treat the world’s biggest diseases. He wondered: How can this process be more efficient?
Very little about drug development is efficient. The failure rate for drugs making their way to commercialization is 90%, after which more than around $1 billion and 10 years is sunk into each one on average.
But technological advancements in data collection are propelling artificial intelligence in drug discovery, which may unlock the ability to find cures for diseases that evaded the scientific community for centuries.
So far this year, startups in drug discovery raised more than $1.4 billion, according to Crunchbase data.
Search less. Close more.
Grow your revenue with all-in-one prospecting solutions powered by the leader in private-company data.
“There are thousands of problems sitting out there that we don’t know the answer for. Thousands and thousands,” said Berlin. “So having a platform that lets us go faster, be precise and scale can really transform the opportunities in front of us.”
How drug discovery currently works
Current processes of drug discovery are long and tedious. Scientists in academia or pharma make molecules. They look for “targets” (like proteins) the molecule can swim to in the body and deliver therapy.
To do that, scientists need to make sure the molecule doesn’t mistake a healthy protein for a target, otherwise a drug swimming around in the body may attach to and kill a healthy cell—amounting to poison. Once scientiststhey get a target, it’s taken out of the body and tested against molecules in the lab to see what will stick.
But as clinical trials continue, several of those drugs fail due to unintended toxicity in the rest of the body, or the drug itself working in the lab but not in humans. With those failures, it sinks millions of dollars and years of research are lost..
“It’s just this huge funnel where stuff can drop out at any point in time,” said Sara Choi, a partner at Wing VC who invests in health startups. “And I think that the problems are very much at the very, very, very beginning of this process.”
Terray and platforms like it work differently. Terray compares molecules against targets, and the AI assesses what parts of the molecule correlate strongly with the target. Terray can then make new molecules that correlate even more strongly, refining it.
Through leveraging data, drug discovery platforms can better predict outcomes of drugs at the start of the process. AI matches molecules with targets and simulates how it will work in the body, giving it a better chance of surviving clinical trials and lowering toxicity rates in patients.
“At the end of the day, it’s about innovation and trying to find interesting, novel ways of treating some really unmet medical needs,” said David Crean, a biotech investor and managing general partner at Cardiff Advisory.
Pharma bets on early-stage technology
AI drug discovery is still nascent, and will require interdisciplinary knowledge of chemistry, computational engineering, machine learning and biology. Data collection in drug development only became popular in 2017, a shift we see in funding: Between 2017 and 2018, funding increased by 190%.
“The foundational layers in terms of data generation were just not there for a long time,” Choi said. “In the last few years we have not seen the breadth of it. We’re just starting a data revolution.”
Nevertheless, large pharma companies like Eli Lilly are betting big on this tech to accelerate the pace of drug development, raising profits and getting medicines into the market faster. Many pharma companies partner with AI drug discovery platforms. For example, Earlier this year, Amgen and Generate Biomedicines announced a partnership potentially worth up to $1.9 billion earlier this yearn.
“The molecules that come out of the drug discovery as a result of AI, there’s only a few in clinical development right now,” Crean said. “It sounds kind of Star Trek-y. Yes, it sounds exciting, but I think we just have to try and manage our expectations.”
Illustration: Dom Guzman
Stay up to date with recent funding rounds, acquisitions, and more with the Crunchbase Daily.