Charles River, Valence Discovery partner on AI-aided drug design
Charles River Laboratories has formed a strategic partnership with Valence Discovery. The collaboration will enable Charles River clients to make use of Valence’s artificial intelligence (AI) platform, which offers machine learning, generative chemistry, and multiparameter optimization.
The Valence AI platform builds on research conducted by the company’s founding researchers at Mila, a large deep-learning research institute in Quebec. The platform enables the design of small-molecule drug candidates, and the following up with rapid optimization against project-specific potency, selectivity, safety, and pharmacology criteria.
Daniel Cohen, Valence Discovery CEO, told Outsourcing-Pharma that the technology benefits from the incorporation of “few-shot” learning, which enables unlocking of prediction tasks that only have small amounts of training data available. This enables users to overcome a significant limitation of existing machine learning technologies in drug discovery.
“The lack of sufficiently large datasets is the most prevalent constraint on deep learning today,” he explained. “Few-shot learning is a class of deep-learning methods that enables us to learn from big data and small data.”
What’s more, he added, the capability to use a smaller amount of data to enact drug discovery puts AI-aided drug design and other advanced tech capabilities in the hands of more pharma professionals.
“By solving some of the key technology challenges in technology-aided drug design, we’re able to bring these capabilities to a larger section of the market than ever possible,” he said.
According to the collaborators, the combination of Valence’s machine-learning technologies with Charles River’s comprehensive capabilities could accelerate discovery efforts at all points in the process. Clients reportedly will be able to benefit from increased diversity in the chemical matter they are investigating, as well more rapid optimization against complex, project-specific design criteria; the combination of capabilities stands to reduce timelines and elevate success rates for drug discovery.
“The Valence platform offers a step-change improvement over existing de novo design technologies,” said Grant Wishart, director of computer-aided drug design with Charles River. “We’ve been consistently impressed by Valence Discovery’s ability to generate high-quality chemical matter that’s readily synthesizable, in novel regions of chemical space, from datasets not otherwise accessible to machine learning methods.”
Cohen added that the agreement with Charles River says something about the value of the Valence technology.
“Charles River is one of the most trusted names across the industry,” Cohen remarked. “The breadth of such partnerships, which span the industry from early to late stages, speak to the power of our platform.”