A significant amount of attention has been given to the design and synthesis of co‐crystals by both industry and academia because of its potential to change a molecule's physicochemical properties. Yet, difficulties arise when searching for adequate combinations of molecules (or coformers) to form co‐crystals, hampering the efficient exploration of the target's solid‐state landscape. This paper reports on the application of a data‐driven co‐crystal prediction method based on two types of artificial neural network models and co‐crystal data present in the Cambridge Structural Database. The models accept pairs of coformers and predict whether a co‐crystal is likely to form. By combining the output of multiple models of both types, our approach shows to have excellent performance on the proposed co‐crystal training and validation sets, and has an estimated accuracy of 80 % for molecules for which previous co‐crystallization data is unavailable.
Cocrystallization is an attractive formulation tool for tuning the physicochemical properties of a compound while not altering its molecular structure and has gained interest from both industry and academia. Although the design strategy for cocrystals has marked several milestones over the past few decades, a holistic approach that utilizes as much cocrystal data as possible is still lacking. In this paper, we describe how information contained in the Cambridge Structural Database (CSD) can be used to construct a data-driven cocrystal prediction method, based on a network of coformers and link-prediction algorithms. Experimental validation of the method leads to the discovery of ten new cocrystal structures for its top ten predictions. The prediction method is not restricted to compounds present in the CSD: by combining the information of only a few cocrystals of an unknown coformer (e.g. an API in development) together with the information contained in the database, a set of relevant cocrystal candidates can be generated.
Cocrystallization has been promoted as an attractive early development tool as it can change the physicochemical properties of a target compound and possibly enable the purification of single enantiomers from racemic compounds. In general, the identification of adequate cocrystallization candidates (or coformers) is troublesome and hampers the exploration of the solid-state landscape. For this reason, several computational tools have been introduced over the last two decades. In this study, cocrystals of Praziquantel (PZQ), an anthelmintic drug used to treat schistosomiasis, are predicted with network-based link prediction and experimentally explored. Single crystals of 12 experimental cocrystal indications were grown and subjected to a structural analysis with single-crystal X-ray diffraction. This case study illustrates the power of the link-prediction approach and its ability to suggest a diverse set of new coformer candidates for a target compound when starting from only a limited number of known cocrystals.
3 December 2018 in Angewandte Chemie International Edition