VantAI, a company focused on generative AI-enabled drug discovery, and Bristol Myers Squibb, a leading pharmaceutical company, have entered into a strategic collaboration to discover new molecular glues for therapeutic targets of interest. The partnership leverages VantAI’s geometric deep learning capabilities with Bristol Myers Squibb’s expertise in targeted protein degradation to discover and develop new small molecule therapeutics.
The collaboration will focus on leveraging VantAI’s generative AI platform to design molecular glues as small molecule therapeutics. These efforts advance VantAI’s mission to unlock the potential of proximity modulation as a powerful tool against therapeutic targets.
VantAI is eligible to receive up to $674 million in discovery, development, clinical, regulatory, and sales milestone payments plus tiered royalties from Bristol Myers Squibb, with an option to further expand to additional therapeutic programs.
“Molecular glues have proven extremely difficult to find, but hold great promise as a treatment modality across a vast array of diseases,” said VantAI’s CEO, Dr. Zachary Carpenter. “At VantAI, we view glue discovery as a challenging ‘geometric puzzle’, and we believe that artificial intelligence is the best tool to find the missing piece. Our collaboration with Bristol Myers Squibb is a significant step forward in our journey to use generative AI to accelerate molecular glue discovery and ultimately bring new therapies to patients.”
“This partnership with VantAI reflects our strategy of leveraging predictive sciences to identify novel molecular glues directed toward biologically validated targets,” said Neil Bence, PhD, Vice President, Head of Oncology Discovery, Bristol Myers Squibb. “We are maximizing the full potential of the our targeted protein degradation research platform based on more than two decades of experience, and collaborating with VantAI will help accelerate our discovery engine and to address critical unmet needs for patients faster.”
VantAI’s technology leverages geometric deep learning to generate insights from millions of years of naturally occurring, evolved interfaces, so that such interfaces can be mimicked during its design process. This ‘Protein-Contact-First’ (PCF) approach simplifies the complex chemical design challenge of bringing proteins together in the cell, yielding non-obvious, more glue-like solutions with optimized parameters including potency, selectivity and molecule size.