To bolster Roche Group’s pioneering efforts in leveraging relational AI for drug discovery, Nvidia has joined forces with Genentech. In a recent announcement, the two companies unveiled a multi-year strategic partnership aimed at combining Genentech’s proprietary designs and data with Nvidia’s cutting-edge AI technology to accelerate AI analysis and expedite the discovery and delivery of innovative solutions to patients.
The centerpiece of this collaboration will be the utilization of Nvidia’s DGX Cloud and BioNemo services, as highlighted in a press release on the companies’ website. Genentech’s researchers will leverage these offerings from Nvidia to scale and optimize their models, leveraging the insights gained to further enhance their product development processes.
Aviv Regev, the Executive Vice President and Head of Genentech Research & Early Development (gRED), emphasized the transformative potential of integrating conceptual AI into their operations. By harnessing Nvidia’s expertise, Genentech aims to streamline the development of therapies that can have a global impact on improving patient outcomes.
Nvidia’s Role in Facilitating Drug Discovery at Genentech
Traditionally, research groups have grappled with the complexities of studying various molecules and their interactions to identify effective treatments for specific diseases, a process that is both challenging and time-intensive. Genentech has recognized the promise of AI in revolutionizing this aspect of drug discovery by swiftly identifying potential drug candidates and interactions through advanced data analysis.
Genentech’s AI/ML teams have already developed state-of-the-art foundational models across diverse research domains, including various medical modalities. Through their collaboration with Nvidia, they aim to leverage BioNemo and the DGX Cloud to further refine these innovative approaches and expedite the drug discovery pipeline.
With access to Nvidia’s DGX Cloud for AI supercomputing capabilities, Roche researchers can enhance their models through pre-training or fine-tuning on specific datasets using BioNemo. Moreover, the integration of BioNemo sky APIs into Genentech’s drug discovery workflows is poised to streamline and optimize their processes.
Nvidia’s technology is poised to accelerate the training and deployment of conceptual models within Genentech’s “lab in the loop” framework, which leverages extensive empirical data to generate predictive models. By validating these predictions in real-world experiments, researchers can iteratively improve their algorithms and develop more efficacious treatments.
While Nvidia’s tools will leverage Genentech’s vast molecular and natural datasets to enhance custom models, access to this proprietary data will be subject to project-specific authorization. Through this collaboration, Nvidia aims to gain valuable insights into AI-driven challenges in drug discovery, informing the ongoing refinement of BioNemo and related solutions to meet the evolving needs of the industry.
