AI3SD Video: Discovery of synthesisable organic materials

The computational discovery of new materials with useful properties is currently hindered by the difficulty in transitioning from a computational prediction to synthetic realisation. Attempts at experimental validation are often time-consuming, expensive, and frequently, the key bottleneck of material discovery.[1] Porous organic cages (POCs) have been discovered as a possible alternative material for molecular separations, catalysis, and sensing applications.[2] For POCs, a priori property prediction is possible,[3] however, it can be time-consuming and computationally expensive to explore a large number of possible candidate molecules. Despite being able to predict materials with exceptional properties, it is often challenging to predict whether it is possible to synthetically realise a potential candidate compound. In the field of drug discovery, machine learning techniques have been able to readily distinguish between synthesisable and unsynthesisable molecules, accelerating the drug discovery process.[4] Incorporating a synthetic accessibility scoring function into the precursor selection process favoured less complex, synthetically accessible precursors; thus, bridging the gap between computational screening and experimental synthesis of POCs. Using data-driven synthetic accessibility scoring techniques and high-throughput experimentation, we developed a POC screening workflow to accelerate discovery of experimentally realisable POC candidates, which we demonstrate using high-throughput, automated experimentation. Existing measures of synthetic accessibility are often tailored towards predicting synthesisable drug-like molecules, whose synthetic requirements often do not align with those of materials discovery programs. By redefining synthetic accessibility as a classification problem, we were able to develop an alternative model able to predict synthesisable materials precursors.[5] Biasing towards easy-to-synthesise precursors facilitated the synthesis of several precursors predicted to form shape-persistent POCs. Using these novel precursors, we were able to construct a precursor library able to be combined using automation, enabling the accelerated discovery of POCs and the construction of an experimentally derived POC reaction dataset. Using this dataset, we aim to develop a model able to predict POC formation, a question challenging to address using conventional computational methods. [1] Szczypiński, F. T.; Bennett, S.; Jelfs, K. E. Can We Predict Materials That Can Be Synthesised? Chem. Sci. 2021, 12 (3), 830–840. [2] Hasell, T.; Cooper, A. I. Porous Organic Cages: Soluble, Modular and Molecular Pores. Nat Rev Mater 2016, 1 (9), 16053. [3] Greenaway, R. L.; Jelfs, K. E. High‐Throughput Approaches for the Discovery of Supramolecular Organic Cages. ChemPlusChem 2020, 85 (8), 1813–1823. [4] C. W. Coley, L. Rogers, W. H. Green and K. F. Jensen, SCScore: Synthetic Complexity Learned from a Reaction Corpus, J. Chem. Inf. Model., 2018, 58, 252–261 [5] Bennett, S., Szczypiński, F.T., Turcani, L., Briggs, M.E., Greenaway, R.L., and Jelfs, K.E. (2021) Materials Precursor Score: Modeling Chemists’ Intuition for the Synthetic Accessibility of Porous Organic Cage Precursors. J. Chem. Inf. Model., 61 (9), 4342–4356