Novartis, Generate:Biomedicines Sign Up-to-$1B AI Protein Drug Collaboration

Novartis will apply Generate:Biomedicines’ generative artificial intelligence (AI) platform to discover and develop protein therapeutics for multiple unspecified disease areas, through a collaboration that could generate more than $1 billion for the developer of therapeutics based on de novo protein generation.

The partners aim to create protein-based drugs more quickly and at lower cost than conventionally developed therapies by combining Generate’s namesake Generate Platform, which integrates machine learning (ML) with high throughput experimental validation, with Novartis’ expertise and capabilities in target biology, biologics development, and clinical development.

Michael Nally, Generate:Biomedicines’ CEO, told GEN Edge the collaboration arose from an ongoing dialogue between his company, based in the Boston suburb of Somerville, MA, and Novartis—which bases its research arm, the Novartis Institutes for BioMedical Research, about 1.5 miles south in Cambridge, MA.

“We’re just right around the corner from a lot of their labs, so we know a lot of the same folks,” Nally said. “Some of the challenges that they were encountering from a protein standpoint, our technology potentially has the opportunity to uniquely answer. We feel like this is a melding of our respective capabilities to try and solve some really important problems for patients.”

Those capabilities, he said, include Novartis’s deep disease-area biology, clinical, and manufacturing expertise, and Generate’s skill in protein molecular design. Generate’s development approach, which it calls “generative biology,” is designed to facilitate testing and learning which proteins work best against which disease targets.

“Our technology, I think, has advanced materially to where we’re seeing a much more robust reproducible infrastructure to both optimize multiple parameters simultaneously, but also find protein-based therapeutics to really difficult targets,” Nally said. “And that’s what’s so exciting for us—it’s to take on some of the problems that the team at Novartis has identified, which traditional technologies struggle to get to good answers on.”

The challenge, Nally explained, was less one of finding targets than to get them druggable enough for further development.

“They have an underlying hypothesis, but they may not have the optimal molecule to prosecute that hypothesis. If you think about where Generate has spent most of its time since inception about six years ago, it’s been in that molecular design space, working across different protein modalities to come up with molecules that are not only fit for purpose for that task, but also have the relevant characteristics of manufacturability and developability, so that when we push these things forward, we find things that could ultimately scale to be important medicine for patients.”

“Generative” biology

Generate’s “generative biology” platform integrates protein science expertise with structural biology, using machine learning algorithms designed to analyze hundreds of millions of known proteins and peptides, looking for statistical patterns linking amino-acid sequence, structure, and function, data science, and computational infrastructure. All that is supplemented with Generate’s own proprietary experimental data.

Generate trains its Generate platform on the entire compendium of protein structures and sequences found in nature, supplementing that data with proprietary experimental data to learn generalizable rules by which a linear amino acid sequence encodes protein structure and function.

Using that knowledge, Generate says, it can create entirely new proteins and modalities that expand its ability to treat disease—a process that according to the company dramatically raises its drug discovery success rate and reduces the time required for drug discovery. The company has projected that its approach can shave two years off traditional protein-based optimization processes.

The Generate platform can also generate antibodies designed to bind specific epitopes on desired targets, allowing in silico generation of potent antibodies on demand. It can also generate functional and agonistic antibodies, as well as antibodies to integral membrane proteins, multiprotein complexes, and other long difficult to hit targets.

While Generate and Novartis say they aim to develop potentially first- and best-in-class molecules through AI-based optimization and de novo generation, the companies are not disclosing the number of targets with which they plan to work, nor the therapeutic areas of the treatments they plan to discover and develop. Generate’s pipeline focuses on three disease areas: immunology, infectious disease, and oncology—while Novartis’ therapeutic areas also include immunology and oncology, as well as neuroscience and cardiovascular, renal and metabolic disorders.

“The beautiful thing about this collaboration is we’re able to tap into their expertise in complementary areas,” Nally explained. “Novartis has an ability to select targets with target-level exclusivity. There is no therapeutic-level exclusivity. So as they look across their research footprint, it’s a matter of finding relevant targets that traditional techniques have struggled to find suitable answers for.”

Modality and disease agnostic

“The technology we’re working on is protein modality agnostic, but also protein disease area agnostic. So, it doesn’t matter whether it’s a neuroscience target or a cardiovascular target, because the technology has applicability in those different domains,” Nally added. “We just need the relevant expertise to come up with the right hypothesis.”

That’s where Novartis is expected to help. In addition to its expertise across drug discovery and development, the pharma giant has years of experience in AI, stretching back to 2019 when it selected Microsoft as its strategic AI and data-science partner for establishing an “AI innovation lab” intended to “transform how medicines are discovered, developed and commercialized.”

“It’s focused on what we call generative chemistry and AI-driven drug discovery, Novartis CEO Vasant (Vas) Narasimhan, MD, said of the Microsoft collaboration at a media event last year. “Our goal now is to invest even more as technology gets better and better within that space.”

“A lot of these natural language processing capabilities could allow us to accelerate and simplify many parts of R&D,” Narasimhan predicted.

In an article posted April 29 on the company’s website, Bülent Kızıltan, PhD, Novartis’ global head of AI & Computational Sciences, declared: “Our ultimate goal is to transform the entire drug discovery process from an AI-enabled to an AI-enhanced and, finally, to an AI-driven process, where the majority of the work will happen in silico.

“This will further empower our medicinal chemists and biologists to focus on elements that require genuine human ingenuity,” Kızıltan added. “This is an extremely exciting prospect.”

$65M upfront

Novartis has agreed to pay Generate $65 million cash upfront—including $15 million toward the purchase of equity in Generate—as well as more than $1 billion in payments tied to achieving milestones. Novartis also agreed to pay Generate tiered royalties up to low double-digits.

The value of Novartis’ collaboration is comparable to another recent Novartis collaboration focused on protein therapeutics.

In April, Novartis inked a worldwide development and commercialization license agreement for Arvinas’ second generation PROTAC® androgen receptor (AR) degrader ARV-766, designed to treat prostate cancer. Novartis also agreed to acquire Arvinas’ preclinical AR-V7 program. In return, Novartis paid Arvinas $150 million upfront and agreed to pay up to $1.01 billion in development, regulatory, and commercial milestones, plus tiered royalties.

Generate:Biomedicines was co-founded in 2018 by a team of leaders from Flagship Pioneering, the venture/accelerator firm whose best-known spinout is Moderna (whose CEO Stéphane Bancel sits on Generate’s board).

Those Flagship leaders included its founder and CEO Noubar Afeyan, PhD, who chairs Generate’s board; Molly Gibson, PhD, an origination partner at Flagship who is Generate’s chief innovation officer; Avak Kahvejian, PhD, a general partner who recently discussed his career on GEN’s “Close to the Edge” interview series; and Geoffrey von Maltzahn, another general partner. A fifth co-founder of Generate, Gevorg Grigoryan, PhD, is the company’s Chief Technology Officer as well as a research associate professor of computer science, biological sciences, and chemistry at Dartmouth College.

Generate emerged from stealth in 2020 when it completed a Series A totaling $50 million, funded solely by Flagship. A year later, Flagship led the company’s $370 million Series B financing, joined by several institutional co-investors.

$693M plus upfront cash raised

Flagship joined numerous other investors (including Amgen and NVIDIA’s venture capital arm NVentures) in Generate:Biomedicines’ $273 million Series C financing completed last year, bringing its total equity financing since 2020 to $693 million, not counting the upfront capital it has garnered through its collaborations.

The sum of that upfront capital is not public, since only the $50 million upfront amount for Generate’s initial collaboration with Amgen has been disclosed. Amgen and Generate agreed in 2022 to discover and create protein therapeutics for five targets across several therapeutic areas and multiple modalities, through a collaboration valued at up to $1.9 billion-plus.

In January, Amgen expanded the collaboration by opting in for a sixth program, agreeing in return to pay Generate an undisclosed upfront payment and up to $370 million in future milestones and royalties up to double digits.

No collaboration values have been furnished for Generate’s strategic collaborations with two top-tier cancer research and patient care institutions.

In April 2023, Generate and The University of Texas MD Anderson Cancer Center agreed to jointly discover and co-develop protein therapeutics for up to five oncology targets in advanced cancers—including small-cell and non-small-cell lung cancer—using the Generate platform. Seven months later, Generate and Roswell Park Comprehensive Cancer Center launched a partnership to discover and develop chimeric antigen receptor (CAR) T-cell therapies, and armoring technologies, for up to three oncology targets, including in ovarian cancer and other solid tumors.

Earlier this year, Generate landed on CNBC’s “Disruptor 50” list of private companies “upending the classic definition of disruption,” where it is ranked No. 25.