A well-funded California biotech startup using AI to design and improve small molecule drug candidates has just struck its third pharma partnership in four years.
Gilead announced a new preclinical research deal with Genesis Therapeutics on Tuesday morning. Genesis, which raised $200 million from investors a year ago, will receive $35 million upfront across three drug targets picked by Gilead, plus undisclosed milestone payments if Gilead takes the drugs into clinical development.
The startup previously partnered with Genentech in 2020 and Eli Lilly in 2022. Evan Feinberg, Genesis’ co-founder and CEO, told Endpoints News in an email that he couldn’t disclose further details about the kinds of diseases or drug targets the company is working on in any of its pharma partnerships.
Gilead is best known for its development of antiviral drugs for hepatitis and HIV, and its CAR-T cell therapy for cancer. While it has dabbled in using AI for development of clinical biomarkers, the Genesis partnership is its first major deal focused on using AI to design drugs.
Genesis launched with a small seed round in 2019, has since raised more than $300 million total, and has “multiple years of runway,” according to Feinberg.
The startup is based around two key AI studies from Stanford University. One described a neural network dubbed PotentialNet to help optimize binding of small molecules to proteins. The second study used AI to predict the absorption, distribution, metabolism, and excretion characteristics of molecules — so called ADME properties — which could help drugmakers pick promising compounds to advance into animal and human studies, and avoid potentially problematic ones.
Genesis has remained quiet about how it’s improved these AI models over the past five years, but has branded its AI platform as GEMS, short for Genesis Exploration of Molecular Space.
“GEMS employs ADME-conditioned language models for molecular generation, diffusion models for protein-ligand docking predictions, and physical machine learning methods for potency prediction,” Feinberg said. He added that GEMS is especially good with “difficult-to-drug targets” and optimizing several parameters of a molecule at once.
The company’s lead program, still in preclinical development, is an inhibitor of PIK3CA, an enzyme with multiple mutations that are linked to cancer. Feinberg said that Genesis is using AI to create a drug that inhibits a range of these mutant proteins while avoiding toxicities from targeting healthy versions of the protein.