Roche’s Genentech unit is partnering with computing giant Nvidia in a multiyear deal that highlights the pharmaceutical industry’s growing hopes of using artificial intelligence to speed up the process of designing and developing new drugs.
By collaborating with Nvidia, Genentech expects to “significantly enhance” its existing AI research and potentially improve on the chip designer’s own software, such as its “BioNemo” platform for drug discovery.
“Our teams will be continuously exchanging expertise on the advancement of science and the state-of-the-art methods emerging in accelerated computing, AI and simulation across this entire drug discovery process,” said Kimberly Powell, vice president of healthcare at Nvidia, in a briefing with reporters.
Successfully developing a new drug from scratch takes a decade or more on average, according to one estimate from the industry group PhRMA. More rapid timelines, such as with messenger RNA vaccines for COVID-19, are often built on years of scientific research that allow drugmakers to bypass sticking points, such as validating a disease target and designing a molecule that has the desired benefit. These phases can take many years, even before a drug candidate reaches human testing.
For years, pharma companies have explored whether AI can shorten the process, striking partnerships with specialists like Exscientia, Recursion, BenevolentAI and Insitro. Interest has remained high even as it’s become clear that available computing technologies don’t solve for all of the biological messiness involved in making a safe and effective medicine. Much-hyped “AI-designed” drugs haven’t yet been the breakthroughs they were initially billed as, for instance.
Genentech is among those drugmakers that have bet heavily on AI. The Roche subsidiary hired computational biologist Aviv Regev from the Broad Institute in 2020 to run research and early development, and in 2021 inked a deal with Recursion to map certain areas of biology using machine learning.
The deal with Nvidia, which is best known for its advanced computing chips, is another bet. The collaboration will use Genentech's proprietary data as well as publicly available data, and focus on early target discovery and molecule development, said John Marioni, senior vice president and head of computation at Genentech, in the briefing.
“I'm confident that by working together appropriately, we can accelerate that process even further, both in terms of biomarker discovery and how we think about designing trials, and that's going to allow us to generate data faster,” Marioni said.
The companies did not reveal specifics about the collaboration’s costs, or identify the disease areas Genentech plans to tackle together with Nvidia. Under the partnership’s terms, Genentech can propose new projects during the course of the collaboration.
Nvidia has significantly expanded its imaging and AI capabilities over the past decade. The company claims its technology can predict protein structures and screen small molecules more accurately and more quickly.
“We know that imaging is at the bedrock of what we do across the healthcare domain,” Powell said. “So whether that is the imaging of cells, organs or radiology, all of that data is how you're going to be able to identify or discover new biomarkers, or you're going to be able to more accurately measure patient response and clinical trials.”
Along with partnerships with companies like Genentech, Nvidia’s venture arm has also been investing in new biotechnology startups. Over the past six months, it’s helped fund Superluminal Medicines, Genesis Therapeutics and Generate Biomedicines.