With $10 million from the Gates Foundation, Schrödinger takes on the toxicity risk

Shortly afterwards, the Schrödinger research team published a commentary in Cell earlier this year, they presented their model predicting binding to the human ether-a-go-go gene (hERG) and sent a copy of it to Bill Gates, whose Bill and Melinda Gates (BMG) Foundation Trust is the company’s second-largest shareholder.

“He was as excited as we thought he would be, like everyone else, and we actually got very, very positive reactions to the paper,” Ramy Farid, President and CEO of Schrödinger, recalled in an interview with GENESIS of the edge.

Among those who responded positively was Trevor Mundel, PhD, who leads the Gates Foundation’s efforts to develop high-impact initiatives against the leading causes of death and disability in developing countries as president of Global Health. Mundel had long been interested in the long-standing biopharmaceutical challenge of predicting toxicity risk early in drug discovery, having approached the FDA when it published its Predictive Toxicology Roadmap in 2017.

“That led to a conversation about how we could work together to help fund the project, and help with discussions with the FDA and discussions with potential partners,” Farid said. “They have a lot of beneficiaries who are trying to develop safe drugs that avoid the need for unnecessary animal testing. So the collaboration was established there.”

These discussions recently led to Schrödinger receiving a $10 million grant from the Gates Foundation to expand a computational platform that can predict toxicity risk early in drug discovery.

Schrödinger said it plans to spend the funding on a computational solution to reduce the risk of development failure associated with off-target protein binding, thereby improving the performance of drug candidates. This involves developing computational methods that can predict how molecules will bind to a panel of off-target proteins—an approach known as predictive toxicology.

Risk Reduction Programs

“The idea of ​​predictive toxicology is to de-risk programs before molecules are given to humans, certainly before they are given to humans, but also before they are given to animals, to predict the toxicity profile or the safety profile of the molecule with respect to binding to extracorporeal targets,” Farid said.

Schrödinger has already used his new structure-based methods to predict binding to hERG, a gene known as KCNH2 encoding protein Kv 11.1, a subunit of the potassium ion channel (I_Cr), which is important for cardiac repolarization. Dysfunction of hERG causes long QT syndrome and sudden death that occur in patients with cardiac ischemia.

In addition to ion channels like hERG, another important class of side targets that Schrödinger is targeting to prevent binding are cytochromes P450. The company has also built structure-based models for cytochrome P450 (CYP) 3A4, 2D6, and 2C9—a trio of side targets that are enzymes in the human liver that metabolize drugs and other substances but that have been linked to adverse drug interactions, increased disease risk, and negative personality traits like anxiety and impulsivity.

Another target that Schrödinger sought to model is the pregnane X receptor (PXR), a key regulator of xenobiotic metabolism and distribution in the liver that can negatively impact drug efficacy and safety when activated by various compounds to speed up metabolism.

“When you’re just working on a project, you have potentially 20,000 proteins in the human proteome that you can bind to. And in many cases, it’s not good to bind to any of them. That’s going to cause some problem that you might not even fully understand,” Farid said.

Farid said Schrödinger’s predictive toxicity solution will involve more than software, potentially including an artificial intelligence (AI) platform and the algorithms that will power it: “It’s existing technology, it’s already proven, it’s already deployed on many of these targets. Our project is industrializing that effort. It’s saying, ‘Look, we’ve got this working on a bunch of non-targets. Let’s do a lot more of them.’”

I’m aiming for 100

For how many non-standard purposes does Schrödinger envisage developing models?

“We would really like to enable and develop a computational toxicology panel in the next few years that includes about 100 secondary targets,” Farid said. “That would be the goal to have all of those models in, say, two to three years.”

Schrödinger’s predictive toxicity integrates physics-based methods and machine learning (ML) to augment these methods. The company says this integration is possible thanks to:

• Its computational platform has achieved a level of accuracy that allows modeling of non-target objects.
• The “revolution” in structural biology is the advancement of experimental and computational methods for predicting protein structures.
• Speeding up computers. Schrödinger uses Nvidia AI technology.

“This is not an ML-only solution. We’ve tried that before. It doesn’t work,” Farid said.

Does Schrödinger envision its predictive toxicology initiative replacing traditional toxicology or the pure machine learning approach used by an increasing number of companies?

“We think it will eventually replace pure ML methods, but it won’t replace ML,” Farid said. “You still need machine learning to be able to scale it up to very large numbers of molecules, while you still need physics to be able to generate a training set. It won’t completely replace experimental methods either. Of course, you still have to test molecules. The idea is to test far fewer molecules experimentally in vitro.”

Dr. Karen Akinsanya, president of Therapeutic Research and Development at Schrödinger, added that the predictive toxicity initiative will not eliminate good laboratory practice (GLP) toxicology studies in animals, which are currently required and regulated by the FDA.

“Today, molecules tend to get pretty far along before we know much about their off-target binding to a lot of proteins,” Akinsanya said. “If you know that really early in the project, it saves you a lot of time and money. You don’t have to get to the end and do all this complicated research and then turn around and start over.”

“Better molecules through discovery”

“This initiative really has the potential to lead to better molecules from these discoveries,” Akinsanya added.

Farid said Schrödinger plans to make its predictive toxin solution available to its customers, “basically the entire biopharmaceutical industry.”

New York-based Schrödinger has been combining forms of artificial intelligence with basic physics principles to identify new drugs for a variety of diseases for about two decades—a key area where the company excels, Farid and Akinsanya explained. GENESIS of the edge last year.

Schrödinger’s activities are based on two pillars:

• Licensing software used in drug discovery and materials design, an activity that has attracted approximately 1,750 customers to the company.
• Implementing its drug discovery platform. Schrödinger has 13 active collaborations with biopharmaceuticals and other partners focused on drug and materials design, with nine of these partners having advanced programs in the clinic.

Schrödinger has discussed its toxicity prediction efforts with several of its biopharmaceutical clients, Farid said, though his company declined to name them: “It’s quite a few large pharma companies. The feedback has been very positive. And that’s unfortunately all we can say.”

Mundel said in a statement that in addition to helping biopharmaceutical companies reduce the time and expense of developing new drugs, predictive toxicity work has the potential to benefit patients in developing countries: “Using computational prediction of the toxicological risk of drug candidates could ultimately improve productivity across the pharmaceutical industry and unlock significant advances in the fight against diseases that continue to plague low- and middle-income countries.”

The $10 million grant to Schrödinger is the third awarded to the company by the Gates Foundation. Schrödinger received $4,938,764 in 2021 to design and synthesize “highly selective and potent Wee2 inhibitors” and another $3,495,888 in 2023 to develop Wee2 inhibitors “that could ultimately lead to a new safe and effective non-hormonal contraceptive.”

Last year, the Gates Foundation gave away $7.7 billion to charities, including nearly $6.3 billion in grants.

The Gates Foundation’s charitable work is funded by the BMG Foundation Trust, which manages funds from Bill Gates, his former wife and philanthropist Melinda French Gates, and Warren E. Buffett, chairman and CEO of Berkshire Hathaway. As of April 1, the Trust owned 11% of Schrödinger’s common stock (6,981,664 shares) and all of the 9,164,193 restricted common shares, according to Schrödinger’s proxy statement filed on April 25.

Eye on Software Revenue Growth

If the predictive toxicology solution initiative is successful, Farid said, Schrödinger expects that success will translate into higher revenue from its software products and services. The company has not publicly predicted when that growth will be expected, although it has acknowledged that it will not contribute to revenue growth this year.

Schrödinger’s software revenue increased by 21% year-on-year in the second quarter to USD 35,404 million from USD 29,352 million in the second quarter of 2023. In the first half of 2024, software revenue increased by 12% to USD 68,819 million from USD 61,565 million in the first three months of 2023.

Successful predictive toxicology activities are also expected to drive the success of Schrödinger’s future drug development programs. Schrödinger enjoyed a doubling of drug discovery revenue in the second quarter, to $11.93 million from $5.837 million in the April-June 2023 period, although the company’s six-month drug discovery revenue of $15.113 million is 61% lower than the $38.406 million reported a year earlier.

“I think the success of this project will definitely lead to increased demand for this technology, which will of course translate into growth in the software industry,” Farid said. “We’re always focused on building technology and science that will have an impact. We have a long history of doing that, and demand will follow.”