Roche, Genentech Commit up to $12B to Recursion for AI-Driven Drug Discovery

Roche, Genentech Commit up to $12B to Recursion for AI-Driven Drug Discovery
Credit: Smokovski/

Digital biology drug discovery company Recursion has secured a commitment of up to $12 billion from Roche and its Genetech subsidiary to apply its Recursion Operating Systemt (OS) to advance therapies across 40 program that include neuroscience and an undisclosed oncology indication the companies announced today.

Recursion OS applies machine learning and high-content screening methods in what the companies said would be a “transformational” model for tech-enabled target and drug discovery.

The integrated, multi-faceted OS is designed to generate, analyze and glean insights from large-scale proprietary biological and chemical datasets—in this case, extensive single-cell perturbation screening data from Roche and Genentech—by integrating wet-lab and dry-lab biology at scale to phenomically capture chemical and genetic alterations in neuroscience-related cell types and select cancer cell lines.

The resulting phenomics data, to be generated in Recursion’s labs, will be analyzed by Recursion’s proprietary convolutional neural networks, which will convert the data into mathematical representations of biology that can be leveraged to identify novel biological relationships and initiate and advance therapeutic programs, the companies said.

Roche, Genentech, and Recursion have also agreed to partner on developing new machine learning algorithms to generate new, “highly granular” maps of human cellular biology.

Those “Phenomaps” are intended to illustrate the inferred relationships amongst perturbation phenotypes in a given cellular context.

Each Phenomap is subject to an initiation fee, acceptance fee, or both that according to Recursion could exceed $250 million for 16 accepted Phenomaps. If Roche exercises its External Use Option for all 12 Phenomaps for which it can do so, Roche’s associated exercise fee payments to Recursion could exceed $250 million.

‘Positive Readthroughs’

“The agreement strengthens RXRX’s balance sheet and has positive readthroughs to its AI/ML [artificial intelligence/machine learning]-enabled platform,” Mani Foroohar MD, Managing Director, Genetic Medicines, and a senior research analyst with SVB Leerink, wrote today in a research note. “While investors will certainly continue to debate the proper valuation and differentiation of RXRX’s analytical platform—both on a standalone basis, and vs competitors and adjacent technologies—we see this transaction as broadly supportive of our thesis that growing demand among large pharma for access to advanced analytical tools for drug discovery will fuel ongoing BD [business development] activity for RXRX.

Roche and Genentech have agreed to pay Recursion $150 million upfront, and up to $300 million in payments toward achieving development, commercialization, and net sales milestones for each of up to 40 programs they have agreed to launch through the collaboration with Recursion, plus tiered royalties on net sales.

“Technology-enabled drug discovery is here, Recursion is leading the space, and we are pursuing some of the most intractable areas of biology with the very best partners by our side,” declared Recursion co-founder & CEO Chris Gibson, PhD, in a statement.

Recursion OS consists of:

  • A network of enabling hardware and software used to designing, executing, aggregating, and storing 11 petabytes of biological and chemical data.
  • A proprietary collection of biological and chemical datasets spanning multiple modalities, including phenomics, orthogonomics, custom assays for program-specific validation, ADMET (absorption, distribution, metabolism, excretion, and toxicology data) assays, and “InVivomics,” consisting of sensor data and continuous video feeds designed to collect holistic measurements of animal behavior and create abstract representations of in vivo disease states.
  • A suite of in-house software applications designed to process and translate data from the Recursion Data Universe into actionable insights.

Up to a decade or longer

Recursion, Roche, and Genentech added that they plan to leverage the insights generated from the collaboration’s maps of human cellular biology to rapidly find and develop medicines against novel targets in neuroscience and the oncology indication for up to a decade or longer.

The collaboration does not include programs already underway at Recursion in oncology or neuroscience, which will continue to be developed independently.

Recursion’s website lists 41 “early development” programs for which details are not disclosed—as well as 15 individual development programs. Of the 15, four have already reached Phase I. The four consist of an oncology program targeting familial adenomatous polyposis (FAP), and programs targeting three neuro disorders: Neurofibromatosis Type 2, cerebral cavernous malformation; and GM2 gangliosidosis.

The other 11 programs in preclinical development target Clostridium difficile colitis; immune checkpoint resistance in  STK11-mutant non-small cell lung cancer; pulmonary arterial hypertension; senolytics; neuroinflammation; small molecule MYC inhibitors; two cancer immunotherapy targets named “Alpha” and “Beta”; hepatocellular carcinoma; Charcot-Marie-tooth disease Type 2; and Batten disease.

Recursion signaled its intent to strengthen its neuroscience focus on November 30, when it announced a trio of new hires designed to deepen its expertise in the field: Tim Ahfeldt, PhD, previously a professor at the Icahn School of Medicine at Mount Sinai, was appointed Fellow, Neuroscience. Irit Rappley, PhD, previously Scientific Director of Discovery and Translational Research, Neuroscience at Bristol Myers Squibb, was named Vice President, Neuroscience and Translational Research. And Glenn Morrison, PhD—a six-year veteran of Genentech and Roche who later held positions at Zogenix and most recently Alector—was tapped to serve as Vice President, Clinical Development.

Roche and Genentech aren’t the only biopharma giants with which Recursion has collaborations. Just yesterday, Recursion said it was expanding its 15-month-old strategic collaboration in fibrosis of the lung, kidney, heart, and other areas, with Bayer to include Recursion’s inferential search capabilities based on its human cellular biology maps—and increase the number of programs from “more than 10” to “more than a dozen.”

The Bayer partnership could also generate more than a billion dollars for Recursion since for each program in development with Bayer, the pharma giant has agreed to pay Recursion more than $100 million in commercial milestone payments plus royalties on future sales. The companies launched their partnership in September 2020, when Bayer’s equity investment arm Leaps by Bayer invested $50 million toward Recursion’s $239 million Series D financing.

In April 2021, Recursion went public through an initial public offering (IPO) that raised net proceeds of $462.6 million. Four months later in August, Bloomberg News reported that SoftBank has built a $5 billion stake in Roche, becoming one of the company’s largest investors, based on its efforts to harness data toward drug development, according to unnamed sources.

“This collaboration highlights the potential of technology to transform drug discovery and unlock previously unknown insights into complex disease in an unbiased way,” added James Sabry, MD, PhD, Global Head of Pharma Partnering at Roche. “We are excited about the opportunity this collaboration offers to help advance the development of medicines at scale.”