Working with sparse data allows us to design first in class molecules for the latest high-interest targets
Generative Design and Active Learning, deliver unprecedented efficiencies to candidate discovery
Balanced molecules encode efficacy, safety and bioavailabilty and maximise the chance of success in a clinical setting
Our systems design molecules that simultaneously address multiple clinical requirements.
Through a combination of AI design, predictive models and experiment, our systems quickly explore and learn which areas of chemistry are most likely to balance the complex requirements for each drug discovery project.
Once promising areas are identified, the systems focus on designs that exploit those areas in order to identify the best compounds.
This balance of early exploration followed by exploitation enables rapid learning and delivers unparalleled progress from initial hits to clinical candidate.
Designing molecules that balance all project requirements of efficacy, selectivity, safety and bioavailability is a significant task.
Generative Design can operate in two and three dimensions. In the following video with use a 3D example to visually demonstrate evolution within a protein structure binding site.
Millions of compounds are evolved during each design cycle and from these we synthesise and test key compounds of highest potential interest. Experimental data not only gives compound specific feedback but is used to assess and refine the predictive power of the model platform.
As a project progresses, the system moves from an exploration phase to an exploitation strategy. By now molecules are consistently fulfilling key project goals and from these we select a candidate molecule suitable for preclinical testing.
