Artificial Intelligence Case Study
Leading generative artificial intelligence (GenAI) algorithms for de novo design are being used routinely within our medicinal chemistry programmes to enable rapid idea generation to address project objectives.
We have successfully deployed Makya, a leading software platform for de novo molecular design from Iktos onto projects to address our client’s needs. These tools use advanced deep learning models to explore vast chemical spaces for new structures that meet multiple design criteria, such as predicted potency, 3D binding poses, and ADMET profiles. This rapid idea generation capability is used widely to solve challenges for our client projects as showcased in the case study below.
Case Study – Application of GenAI to the design of CNS-penetrant leads on a neurology project.
A client project focused on a CNS indication required a multi-parameter optimization (MPO) strategy to enhance CNS penetration. To support this, Generative AI was used to propose modifications at the R1 position, resulting in a remarkably diverse and novel set of 27,000 ideas, generated within just a few hours. These ideas were evaluated using a custom, project-specific KNIME-Hub SaaS workflow that modelled ADME properties across multiple parameters, to enable efficient triage. Hundreds of candidates were reviewed collaboratively by the Computational Chemistry and Medicinal Chemistry teams. This multidisciplinary effort led to the prioritization of 5 compounds for synthesis. Of these, Compound Idea 818 demonstrated the desired profile; strong binding affinity, favourable ADME characteristics, and good brain penetration (Kp,uu0.6).
- Fast idea generation: 1000’s in just a few hours of computation, sampling a massive chemical space
- Chemically tractable ideas with amenable reactions
- ML models for MPO aid prioritisation, allowing multidisciplinary team to select suitable synthesis targets
- Achieved project goal with improved Neuro profile