AI drug development company Genesis Therapeutics recently announced that it has received additional investment from Nvidia's venture capital arm, NVentures, with the specific amount undisclosed. This move signifies a deepening of the collaboration between the two parties, aimed at accelerating the development of Genesis's AI platform GEMS (Genesis Exploration of Molecular Space), which focuses on structure-driven drug design using physics-based AI.
Originating from Stanford, Deepening Molecular AI
Genesis Therapeutics was derived from Dr. Vijay Pande's lab at Stanford University. Co-founder Dr. Evan Feinberg co-invented and co-authored several key papers on deep learning technologies with Pande during his graduate studies, notably the PotentialNet algorithm. This algorithm was the first to use a novel graph neural network for predicting molecular properties, particularly protein-ligand binding affinity. Feinberg, Pande, and their colleagues demonstrated the performance of PotentialNet in potency prediction and further validated its effectiveness through collaboration between Stanford University and Merck Research Laboratories. Before founding Genesis, Feinberg served as a deep learning consultant at Merck.
Image Source Note: Image generated by AI, image authorized by service provider Midjourney
Over $300 Million in Financing, Deep Collaboration with Nvidia
Genesis was founded in 2019 and secured $52 million in Series A funding a year later. Since then, the company has continued to grow, raising over $300 million to date, with most of this funding coming from the $200 million Series B financing completed in 2023, which included investments from Nvidia's venture capital arm NVentures.
Through its collaboration with Nvidia, Genesis is working to accelerate the development of its AI platform GEMS. GEMS aims to generate and optimize molecules for complex targets by integrating proprietary AI methods, including language models, diffusion models, and physics-based machine learning (ML) simulations. The additional funding from NVentures is intended to enhance computational efficiency by leveraging Nvidia's expertise, further enhancing Genesis's physics-based AI platform for structure-driven drug design.
Feinberg stated, "Nvidia is a leader in many aspects of the AI stack, both in hardware and in the lower software layers above the hardware. Genesis has always aimed to be a pioneer in the molecular AI space. Therefore, there is a very clear synergy between Nvidia's comparative advantages and Genesis's comparative advantages, making the combined power greater than the sum of its parts."
Optimizing Neural Networks to Accelerate Drug Development
The collaboration will cover the optimization of equivariant neural networks, which are particularly valuable for processing 3D geometric data such as protein and small molecule structures. Nvidia has been committed to accelerating computation through neural networks, including training networks and running inference, using trained models to predict new data or deploy in real-world environments.
Feinberg explained, "For the molecular AI field that Genesis has been pioneering for years, there are specific types of neural networks that are particularly useful. This is essentially a continuation of a long-term trend in the field, where AI is not a monolith. There are many subfields of AI, each using related but different algorithms for learning."
At Stanford, Feinberg, Pande, and a group of colleagues introduced the PotentialNet graph convolution family in a paper published in ACS Central Science in 2018. Two years later, another group of colleagues, along with Feinberg and Pande, demonstrated how to achieve "unprecedented accuracy, to our knowledge," in predicting ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties by explicitly representing each molecule as a graph, showcasing the significant advantages of AI algorithms over advanced ML used by Merck Research Laboratories in a paper published in the Journal of Medicinal Chemistry.
Close Collaboration Between Founder and Mentor
Pande is now a general partner at Andreessen Horowitz (a16z) and a founding partner of the a16z bio fund, leading the firm's investments in biology, computer science, and engineering. Pande was Feinberg's doctoral advisor and led a16z's $4.1 million seed investment in Genesis, co-leading the company's over $200 million Series B financing with an undisclosed U.S.-based life sciences investment firm.
Feinberg spoke highly of Pande, saying, "I feel very fortunate to have collaborated with him for nearly a decade. I think it's rare to work so closely with someone so talented and visionary and to learn from him."
Continuous Innovation, Leading Industry Development
Feinberg added, "He (Pande) has always pushed me in ways that are crucial to Genesis's success. As the field evolves, he continues to evolve. I think this aligns with how we maintain our leadership position in the field, where we continuously innovate rather than merely being satisfied with imitation, but truly driving the field forward."
Feinberg recalled that during his graduate studies at Stanford, AI primarily impacted computer vision and natural language fields. "The types of neural networks used for these two were actually quite different from each other and not very applicable to chemistry. So, we developed new types of neural networks," Feinberg recalled, "In the mid-2010s, graph neural networks became more suitable for molecules."
Feinberg stated that since then, Genesis has continuously researched new AI algorithms, "new types of neural network primitives that are more suitable for molecular AI tasks." "Equivariant neural networks are one of the series we value. This is also one of the areas where Nvidia is particularly helping us optimize," Feinberg added.
Pande's lab initially gained fame for its distributed computing project Folding@Home, which aimed to simulate protein dynamics, including the protein folding process.
Feinberg recalled, "Folding@Home leveraged a vast number of Nvidia GPUs globally for protein folding simulations. Since then, Nvidia GPUs have been increasingly used for AI, especially in vision and natural language. Therefore, we can say our company is a powerful user of Nvidia GPUs."
A "Perfect Match" with Nvidia
Feinberg remarked, "When we were introduced to Nvidia and NVentures through our Series B financing, it felt like a very natural investor who would not only bring a lot of capital but also wisdom to the relationship. This investment truly laid the foundation for our collaboration beyond a customer relationship, facilitating our mutual learning from our needs and their lower-layer capabilities that we can uniquely leverage from our domain knowledge."
For Nvidia, collaborating with Genesis strengthens its ongoing efforts to apply AI to drug discovery.
Nvidia's Vice President and head of NVentures, Mohamed "Sid" Siddeek, stated, "Genesis's AI platform and the related computational advancements developed in collaboration with Nvidia will help provide new generative and predictive AI technologies to explore untapped chemical pathways and identify drug candidates."
How Does GEMS Assist Nvidia?
Feinberg stated, "The goal of GEMS is to be able to effectively develop very challenging targets that, in some cases, may even be undruggable. To do this, we need to perform several capabilities better than before."
This includes generating molecules and predicting their potency, selectivity, and atomic properties—a unified multi-parameter optimization approach for jointly studying all key characteristics of molecules in drug discovery. Feinberg explained that GEMS consists of two deeply integrated pillars—generative AI and predictive AI—and has generated thousands to millions, even billions, of compounds in the cloud using Genesis's own custom language models.
"But chemistry, synthetic chemistry, is the limiting factor. Only so many molecules can be made in a given time. Therefore, our predictive AI technology (for predicting potency, selectivity, and atomic properties) is crucial to being as accurate as possible. Thus, GEMS is essentially a term that describes a deep integration of technological combinations," Feinberg said.
Applications of GEMS in Oncology and Immunology
Using GEMS, Genesis is developing a pipeline focused on oncology and immunology. In oncology, Genesis is in the late lead optimization phase, nearing the nomination of what it calls a highly potent and selective development candidate for the PIK3CA pan-mutant allosteric inhibitor, a common oncogenic driver in breast and colorectal cancer.
Other oncology development efforts focus on small molecules aimed at overcoming responses to checkpoint inhibitors (lead optimization phase) and anti-apoptotic modulators to prevent cancer cells from evading apoptosis by inhibiting exogenous cell death pathways (discovery phase).
In immunology, Genesis reported that it has two discovery phase projects: one aimed at developing multiple small molecules targeting well-validated autoimmune disease targets; the other using small molecule correctors to restore the activity of undisclosed impaired proteins to treat "severe hereditary autoinflammatory diseases."
Collaborations with Biopharmaceutical Giants
In addition to internal development work, Genesis is also collaborating with three biopharmaceutical giants, although Feinberg stated the company cannot comment on this. The most recent collaboration was initiated in September with Gilead Sciences, which agreed to use GEMS to assist in generating and optimizing molecules for Gilead's selected targets, thereby discovering and developing small molecule therapies for multiple targets.
Gilead agreed to pay $35 million for three targets and has the right to nominate additional targets at an undisclosed pre-agreed fee per target. Gilead also agreed to make additional payments related to achieving preclinical, development, regulatory, and commercial milestones, as well as tiered royalties on net sales of commercialized products.
Collaborations with the other two biopharmaceutical giants include:
- Eli Lilly - a collaboration valued at up to $670 million (including a $20 million upfront payment), aimed at discovering new therapies in up to five therapeutic areas, launched in 2022.
- Genentech, a member of Roche Group - a collaboration involving multiple targets and diseases, launched in 2020, utilizing Genesis's platform for deep learning and molecular simulation. In 2022, Genentech described its targets of interest as "challenging targets that cannot be reached by other methods." The value of this collaboration has not been disclosed.
Genesis is headquartered in Burlingame, California, in the San Francisco Bay Area, with a fully integrated lab in San Diego. The company employs around 80 people.
Feinberg stated, "We do have a significant expected growth, driven in part by the Series B financing, Nvidia's latest investment, and our partnerships. I don't have an exact number on how large we will be in 12 months, but we do have enough scale to surpass 80 people."