BIOBRIEF
Recursion and Genentech's AI drug discovery partnership produced its first confirmed drug target in neuroscience — a gene or protein the labs agreed to pursue — after 15 months. The target, mechanism, and disorder remain undisclosed.
Neuroscience drugs have historically failed more than 9 times out of 10 once they enter Phase I clinical testing. On Thursday, Recursion, a Salt Lake City-based AI drug discovery company, and Genentech, the Roche subsidiary, exercised the first option in their 4.5-year, up-to-$12 billion partnership: a validated neuroscience target produced after a 15-month validation cycle. The target itself, its mechanism, and the disorder it might address remain undisclosed.
A validated target is a gene or protein the partner labs have agreed is real and worth pursuing as a drug target. Genentech's acceptance of Recursion's target validation package triggers a $3 million milestone payment, lifting cumulative cash paid to Recursion under the 2021 collaboration to $216 million. The SEC-filed collaboration agreement sets a deal ceiling of up to $12 billion spread across as many as 40 programs, with per-program milestones up to $300 million plus tiered royalties.
The field's structural difficulty is the yardstick. Per the BIO/BioMedTracker/Amplion Clinical Development Success Rates 2006–2015 study, only about 8.4% of neurology candidates entering Phase I reach approval. Recursion frames the milestone as evidence that its Recursion OS platform, fed by Roche and Genentech's single-cell perturbation data, can surface novel neuroscience targets where conventional discovery has plateaued. One validated target against a 40-program sample is the first measurable input into that test. The next checkpoint is preclinical entry, the IND-enabling work that precedes a Phase I trial.