Model Medicines Publishes Groundbreaking Preprint on AI-Driven Drug Discovery
Model Medicines is proud to announce the publication of our first preprint, "ChemPrint: An AI-Driven Framework for Enhanced Drug Discovery," now available on bioRxiv. This preprint showcases the groundbreaking work being conducted by our team, leveraging the power of artificial intelligence to revolutionize the drug discovery process.
Highlights
Unprecedented Efficiency: ChemPrint, our proprietary AI model, achieved a remarkable 45.5% in vitro hit rate in our oncology program targeting AXL and BRD4, a massive increase over average rates reported by industry leaders like Schrödinger and others.
Novel Chemical Discoveries: Compounds identified by ChemPrint exhibited substantial chemical novelty, with an average Tanimoto similarity score of 0.32 to their training set, significantly diverging from known compounds and demonstrating ChemPrint's capability to extrapolate beyond traditional benchmarks.
Innovative Methodology: Our novel train-test splitting methodology, utilizing t-distributed Stochastic Neighbor Embedding (t-SNE), enables our models to navigate uncharted molecular territories accurately.
Adaptive Molecular Embeddings: ChemPrint employs adaptive molecular embeddings that preserve vital chemical and structural information throughout the AI discovery process, facilitating a more nuanced comprehension of the molecular interactions fundamental to identifying viable therapeutic agents.
The GALILEO™ Advantage:
This preprint underscores the capabilities of our GALILEO™ AI drug discovery platform, which incorporates ChemPrint as its foundational model. By leveraging adaptive molecular embeddings and stringent model training environments, GALILEO™ improves the efficiency of drug discovery, addressing the challenges of low hit rates and the difficulty in exploring novel chemical spaces.
BioRxiv - “ChemPrint: An AI-Driven Framework for Enhanced Drug Discovery”
LEAD AUTHORS
Tyler Umansky, Navya Ramesh, Virgil Woods, Sean M. Russell, Davey Smith, Daniel Haders, Model Medicines, University of California San Diego
