Schrödinger’s just-launched, potentially more than $2 billion collaboration with Novartis is the 20th to be listed in its partnered pipeline, and the first one involving a longtime customer in Novartis, in an example of a business development call paying off.
The pharma giant has inked a research collaboration and license agreement with Schrödinger to advance “multiple” small molecule candidates into Novartis’ portfolio for further development.
In addition, Novartis agreed to a three-year expanded software agreement that increases its access to Schrödinger’s computational predictive modeling technology and enterprise informatics platform. The value of that expanded agreement was not disclosed.
Novartis had already been using Schrödinger’s software in drug discovery for “likely decades,” Schrödinger president and CEO Ramy Farid, PhD, estimated, when his company reached out to its longtime customer.
“We had the opportunity at some point in the year to describe some of our undisclosed programs that we were working on, and they expressed interest in them. They shared with us some of the programs they’re interested in and targets they’re interested in. They were of mutual interest,” Farid told GEN Edge.
Karen Akinsanya, PhD, Schrödinger’s president of R&D, therapeutics, told GEN Edge Novartis’ interest in a collaboration grew as the companies talked.
“We were trying to show them exactly how we use the platform: What is the workflow? How do we identify structures, refine structures, find binding sites, validate binding sites, all the steps that it takes to get ready for our platform? We used some examples from our portfolio to demonstrate that,” Akinsanya recalled. “And of course, they were very excited about perhaps bringing that workflow inside.”
Novartis has agreed to scale up its use of Schrödinger’s software across its global research organization for faster, more intense use, Farid said, by increasing the number of licenses for the technology and performing more calculations.
As a result, the companies said, Novartis will be able to use Schrödinger’s full suite of drug discovery technologies across its research sites, with Schrödinger agreeing to provide support comprehensive enough to ensure full integration and optimization of the platform.
“Enough” licenses
The scale-up is significant, Farid explained, since Schrödinger licenses its computational predictive modeling software by usage and its enterprise informatics system by number of users. The latter stores experimental and computational data and allows medicinal chemists, modelers, and other drug discovery team members to collaborate.
“Novartis will have enough licenses now to be able to deploy LiveDesign [digital molecular design and discovery platform] through its enterprise informatics system to their broader research teams, not just to a small number of people,” Farid said.
As for the computational predictive modeling software, he continued, “They will be able to calculate the properties of molecules more rapidly because now they have more licenses.”
Those properties, Farid said, can include levels of potency, selectivity, solubility, and permeability for a given target, among other properties.
“There’s something else that’s more important: They can now compute the properties of more molecules,” Farid added. “That increases the probability of identifying a molecule with the right properties, because the more molecules you test computationally, the more likely you are to find a molecule that has just the right properties.”
While artificial intelligence (AI) and machine learning are components of the predictive modeling platform, Schrödinger said the platform is better described as being physics-based but AI-enhanced.
“What we do is we can build training sets to train AI and machine learning, but on a scale that totally dwarfs experiment,” Farid said. “You can compute the properties of tens of thousands of molecules in a day. That would take years to do experimentally. We can do it in a day and obviously much more cheaply using physics. And the reason you can use our physics is because it’s as accurate as doing the experiment. That’s how you combine physics and machine learning.”
Based in New York, Schrödinger has combined forms of AI with physics-based first principles for about two decades, in order to identify new drugs for targets that are designed to treat a variety of diseases—a key area where the company stands out, Farid and Akinsanya explained to GEN Edge last year.
Schrödinger’s business consists of two prongs:
• Licensing its software used in drug discovery and materials design, a business that has attracted more than 1,750 customers to the company.
• Deploying its platform to drug discovery. Schrödinger maintains approximately 20 active collaboration projects with biopharma and other partners focused on drug and materials design, with nine of those partners having advanced programs into the clinic.
Undisclosed targets
The companies plan to combine their research efforts to identify and advance into programs therapeutics that Novartis has agreed to develop and commercialize. The therapeutics will be designed for undisclosed targets in Novartis’s core therapeutic areas.
Novartis’ website lists its therapeutic areas of focus as cardiovascular, renal, and metabolic disorders; immunology; neuroscience; and oncology. Schrödinger lists 11 oncology candidates, and two candidates each in immunology, neurology, and central nervous system (CNS) in its proprietary and partnered pipelines.
The companies aren’t disclosing which therapeutic areas they will focus on.
The therapeutic areas common to both companies will not necessarily be the ones to be addressed through the collaboration, Farid said, because the programs Novartis has licensed for co-discovery with Schrödinger include non-oncology discovery candidates listed in Schrödinger’s pipeline that have not been disclosed.
Two of Schrödinger’s eight disclosed proprietary candidates are discovery phase programs—one targeting NLRP3 in immunology and the other targeting LRRK2 in neurology. The pipeline also includes an undisclosed number of unspecified programs in “multiple areas.”
Novartis agreed to pay Schrödinger $150 million upfront and up to $2.272 billion in total payments tied to achieving milestones across an initial set of programs. The payments will consist of up to $892 million in discovery and development milestones, and up to $1.38 billion in commercial milestones. In addition, Schrödinger is entitled to a tiered percentage royalty ranging from mid-single digits to low double-digits on products commercialized by Novartis under the agreement, subject to specified reductions.
The $150 million upfront is more than four times the $35.29 million that Schrödinger generated in total third-quarter revenue. Nearly all of that (90%) was the $31.884 million generated in software products and services, with the remaining 10% ($3.406 million was drug discovery revenue.
Stock surge
Investors responded to news of the Novartis collaboration on November 12 with a buying surge that sent Schrödinger’s share price on NASDAQ jumping 14%, from $19.54 to $22.25. Novartis shares traded on the SIX Swiss Exchange dipped nearly 1% from CHF 93.09 ($105.13) to CHF 92.36 ($104.31).
The collaboration with Schrödinger marks Novartis’ latest in a flurry of drug discovery partnerships launched in recent weeks. On Monday, Novartis committed up to $745 million plus potential royalties to partner with Ratio Therapeutics on developing a Somatostatin Receptor 2 (SSTR2) radiotherapeutic candidate for cancer.
Last month, Novartis agreed to pay Monte Rosa Therapeutics $150 million upfront toward a potentially more than $2.1 billion collaboration to develop T and B cell-modulating molecular glue degraders targeting the VAV1 protein in multiple therapeutic areas, starting with its Phase I autoimmune disease candidate MRT-6160.
And in September, Novartis paid Generate:Biomedicines $65 million upfront to launch a potentially more than $1 billion partnership to discover and develop protein therapeutics for multiple unspecified disease areas, using Generate’s generative AI platform.
“We are excited to build on our long-standing relationship with Schrödinger, leveraging their discovery platform and physics-based computational methods to accelerate our drug discovery efforts,” Fiona Marshall, president of biomedical research at Novartis, said in a statement. “These agreements underscore our commitment to innovative computational technologies that enhance our research capabilities to help us bring new and impactful medicines forward faster.”
Below analyst forecasts
Schrödinger’s stock surge came despite the company reporting third-quarter revenue that fell below analyst forecasts.
Schrödinger’s Q3 revenue fell 17% from the year-ago quarter. That reflected a 75% year-over-year decline in drug discovery revenue since the $13.665 million generated in Q3 2023 reflected accelerated recognition of deferred revenue associated with programs no longer in the company’s collaborative portfolio. The company scaled down its drug discovery revenue guidance range to between $20–$30 million, from $30–$35 million.
Software revenue, by contrast, rose 10% from $28.904 million in Q3 2023. The net loss of $38.136 million was 29% less than the $62.024 million net loss reported a year earlier. And Schrödinger increased the lower end of its software revenue growth guidance range for 2024, from 6–13% to between 8–13%.
“The miss was not anything actually occurring in the business during the quarter, it was just timing of certain events” changing from before to after the end of the quarter, Farid explained. “I think that’s the reason why the focus was able to remain on the Novartis collaboration and the excitement around, obviously, the potential value that’s going to be generated for the company.”
That value starts with the $150 million upfront that Novartis has agreed to pay Schrödinger.
“That’s a very meaningful upfront for a discovery program, and that should not be missed,” Farid said. “It sends a really clear message that unlike a lot of the AI deals that people are seeing, where the upfronts aren’t even disclosed, this is a reflection of the track record that we have, the efficacy of the platform, and our track record of being able to deliver development candidates.”
