The 4.5 year, up-to-$12 billion artificial intelligence (AI) drug discovery collaboration of Recursion, Roche, and its Genentech subsidiary marked a milestone Wednesday when plans were announced for Recursion and Genentech to co-develop a neuroscience small molecule early discovery program based on the first validated target discovered through an AI map they created to find new targets in a long-challenging therapeutic area.
Genentech has exercised the collaboration’s first validated target option after accepting from Recursion a validation package for the target. That acceptance will trigger a $3 million milestone payment from the biopharma giants to Recursion. That latest payment raises to $216 million the total cash paid out by the biopharma giants to Recursion since the AI-based drug developer joined with the pharma giant and its subsidiary to launch their partnership in 2021.
For each program developed, Recursion can receive up to $300 million in payments tied to achieving development, commercialization, and net sales milestones, as well as tiered royalties up to high single digits per small molecule program. Roche and Genentech have committed to using Recursion’s platform to advance therapies in 40 programs that include “key areas” of neuroscience and an undisclosed gastrointestinal (GI)-oncology indication.
The companies are not disclosing what the target is, or what neuroscience disorders it has the potential to treat. Recursion does say, however, that it took the companies 15 months to go from initiation of target validation to a validation package.
The target’s significance, according to Recursion and its partners, rests in its potential to expand the universe of neuroscience treatments beyond the handful of familiar targets in the space.
While more than three billion people are affected by neurological conditions worldwide, only one in 40 neuroscience drugs to ever reach the clinic have been approved. Just 8.4% of neurology drug candidates that enter Phase I studies reach all the way to approval, according to a Clinical Development Success Rates 2006-2015, a study published by the Biotechnology Innovation Organization (BIO), BioMedTracker, and Amplion (acquired in 2022 by Science and Medicine Group or SMG).
“Finding new targets in neuroscience has historically been challenging, and this milestone highlights our ability to uncover novel biology in areas where conventional approaches have struggled,” Recursion CEO Najat Khan, PhD, said in a statement.
New tools
New hope for discovering and developing new neuroscience drugs, the companies say, is a result of new tools developed to study the mostly unexplored genome in living neurons.
Those new tools start with Recursion’s Data Factory, an end-to-end platform developed and perfected for more than a decade to create a repeatable, scalable system for generating biological data designed for AI models.
“We really took an approach to be unbiased in our full genome-wide scan of these opportunities to uncover new biology,” Christopher Winrow, PhD, Recursion’s vice president of neuroscience, told GEN.
Partnering with Roche and Genentech, Recursion built the first whole-genome CRISPR knockout map generated from a subset of over one trillion internally manufactured neuronal cells derived from induced pluripotent stem cells (iPSCs)—about 12 brains’ worth of neurons—after outside cell manufacturers told the partners that the scale of production they were seeking was too difficult and too expensive to carry out.
“The cell context is important. We’re starting with human iPSCs and driving these into a very clearly homogenous population of neurons that we can test—and we do this at scale,” Winrow said. “You need a certain differentiation period for the iPSCs to form into the neurons that you want to study. There’s also a certain QC [quality control] that you need to do to make sure that those neurons are what we expect them to be, and that that’s robust and reproducible.”
Recursion subjected the potential targets they identified to what the company said was a rigorous validation process developed jointly with Genentech. Candidate targets advanced through successive stages of pathway validation, functional validation, and disease validation to determine whether modulating the target altered neurological disease phenotype. Only targets that consistently showed compelling evidence across
each stage advanced into a validation package.
Researchers from Recursion and Genentech applied a large-scale perturbation set using whole-genome CRISPR-Cas9 knockouts spanning more than 17,000 genes and thousands of small molecules. “You look for what we would call gene-compound interactions, to uncover some of those insights, at the same time as looking at the gene-gene interactions,” Winrow said.
“We’re looking at whole genome-wide knockout, not just a handful of areas or pathways of interest. That’s really a big game changer, in that we have this broad view,” he added. “As we perturb the cells, we start to see the known actors. But more importantly, we’re starting to see these unexplored areas really come to light. So that’s what’s changed, taking that broad view at a genome-wide level with the scale and throughput that we have at Recursion.”
46M+ images
With the cells produced and perturbations added in, the next step entails imaging designed to capture more than 46 million cellular images. Each image is analyzed across hundreds of features such as mitochondrial shape, nuclear morphology, using AI foundational models.
“What even is more mind-blowing beyond that is there are multiple features within each of those images,” Winrow said. “Traditionally as a scientist, I’d go in and I’m really interested in mitochondria. So, I look at the image and I say ‘Wow, am I seeing mitochondrial fragmentation?’ Well, that’s great. Yes, I am, right?’ But if you’re looking at hundreds and hundreds of different features, that could be a mitochondrial shape, it could be nuclear capacity, it could be a whole bunch of different things that the AI models are trained upon, that then enables a real richness to come out of that dataset.”
“So, it’s not just the scale of those images, it’s actually what’s within those images that we require the AI approaches to really uncover,” Winrow added.
The resulting data is analyzed to identify patterns resembling disease biology using Recursion’s BioHive-2 supercomputer, completed in 2024 based on Nvidia technology and proprietary AI models. Those insights are then turned into hypotheses that govern further research and ultimately, validated targets.
“Each target is individual. They are related to a core disease process,” Winrow said. “We have these core areas of biology that are looked at by a lot of different groups. We want to understand the unexplored biology around those areas, and that can be anything. It’s an unbiased approach, so all of those things that are associated have not necessarily been connected to this target in the past, so you come up with all sorts of different targets.
“Validation, then, is central to that core biology in exploring a whole bunch of different new intersecting targets,” he added.
Next steps
Next steps for the partners include advancing the discovery program from a target into a drug, using Recursion’s chemistry platform to design a potential first-in-class molecule—as well as identifying and validating additional targets for potential new programs.
Recursion said it will continue to combine its phenomics dataset with Genentech’s proprietary transcriptomics data to build additional multimodal maps designed to explore potential novel targets and pathways by systematically linking gene perturbations to cellular phenotypes.
The neuronal map is one of two whole genome neuroscience phenomaps produced by the Recursion-Roche-Genentech partnership. The other is a map of specialized microglial immune cells unveiled last year, which the companies also plan to use toward revealing significant new targets in neurodegenerative diseases.
“This provides us really two distinct but CNS-focused maps that are rooted in human biology. I think the opportunities are there to really mine both of those. The neuro map is a little bit ahead, and so we’ve been able to take some of the learnings that we gained there and apply that to our work on the microglia side as well.”
The two neuro-focused maps, in turn, are among six total whole genome phenomaps produced by the companies; the other four are designed to discover targets for a single undisclosed indication in GI oncology. The maps are designed to allow Recursion, Roche, and Genentech to explore potential undiscovered targets and pathways, systematically linking gene perturbations to cellular phenotypes.
“Because we can reuse these maps again and again, I might find something in the microglia map that’s really intriguing that I can now follow up in in the neuron map, and vice versa,” Winrow commented. “I think that’s a real power of this approach as well.”
The post Recursion Partners with Genentech to Advance First Validated Neuro Target Discovered Through AI Map appeared first on GEN – Genetic Engineering and Biotechnology News.
