Generative Design of Selective ADORA3 Agonists for an Anticancer Target
Case Study 03
Machine learning and generative design were used to create selective adenosine A3 receptor agonists, balancing target activity with drug-like properties, synthesizability, and patent novelty.
Challenge
The project focused on designing new molecules for the adenosine A3 receptor (ADORA3), an anticancer target, while maintaining selectivity over the related ADORA1, ADORA2A, and ADORA2B receptors.
What We Did
-
1
Built and validated receptor models
Curated activity data for ADORA3 and the related adenosine receptors, then generated and validated machine learning models to guide selective design.
-
2
Generated new candidate molecules
Used MegaSyn with known agonists and the receptor models to design new molecules, alongside ADME models for properties including metabolic stability, solubility, and permeability.
-
3
Assessed synthesis and patent novelty
Applied in-house synthesizability and retrosynthesis predictions, and compared designs with published patents using SureChEMBL to inform freedom-to-operate assessment.
-
4
Selected a compound for synthesis
Worked with the client to select a compound for synthesis at CPI.