DNA-encoded library screening provides rich datasets of billions of observations for a target screened under multiple, biologically relevant conditions.
These data are too large and complex for a human to interpret yet embedded within, are the roadmaps to making better products. Machine Learning (ML) models built upon these large datasets help us uncover key properties of the target we want to exploit as well as properties we want to avoid. Our state-of-the-art ML pipelines rapidly build powerful and insightful models with these data.
We then use trained ML models to search virtual chemical space and to support compound optimization. By using these models in tandem with scientists’ insights, we can efficiently discover potent and selective molecules with favorable properties.