CAMBRIDGE, UK, 2 July 2024 – Optibrium, a leading developer of software and AI solutions for molecular design today announced the publication of a peer-reviewed study in the Journal of Computer-Aided Molecular Design, ‘From UK-2A to florylpicoxamid: Active learning to identify a mimic of a macrocyclic natural product’. The paper demonstrates the successful application of the QuanSA (Quantitative Surface-field Analysis) method, part of Optibrium’s BioPharmics platform for 3D molecular design, to accelerate the lead optimisation of a complex macrocyclic natural product during agrochemical development. By significantly reducing the number of synthetic steps required during optimisation, the study supports the commercial viability of complex macrocyclic compounds.

Optibrium’s QuanSA method uses an active learning approach that combines two types of molecular selection—the first identifies compounds predicted to be most active, and the second identifies compounds predicted to be most informative for lead optimisation. The method has broad applications in lead optimisation where scaffold replacements are needed, from agrochemical development to small molecule and macrocyclic ligand design and discovery. In the study together with a leading agriscience company, Optibrium explored how this approach could provide a more efficient route to finding new agrochemicals (e.g., for crop protection) by reducing the number of compounds requiring synthesis.

Florylpicoxamid (FPX) is a mimic of a macrocyclic natural product, UK-2A, originally identified through a stepwise deconstruction method that required thousands of synthetic analogues alongside in vitro and in planta assays. Using the QuanSA method, the binding metabolic form of FPX was successfully identified within five rounds of compound selection and model refinement, reducing the total number of required synthetic analogues by a factor of ten.

Purely ligand-based affinity prediction is challenging, with the presence of macrocycles compounding the complexity. We are excited to show how machine learning can build physically meaningful models for lead optimisation and how Optibrium’s QuanSA method, using an active learning strategy, can be applied to real-world molecular design.

Ann Cleves, VP of Application Science, Optibrium

Ann continues: “Macrocyclic natural products show great promise as drugs and in crop protection, but their complexity makes them difficult to synthesise and implement on a large-scale. This study demonstrates that we can greatly simplify the lead optimisation of complex molecules not only for drug discovery but to drive new agrochemical development.”

Related content from across the site

Meet with us at ACS Fall 2024

Join Daniel Barr to hear more about how deep learning imputation prioritises the most relevant data, accounts for uncertainty, and guides experiment selection to bring additional value to small molecule discovery.