MS31-P05 Discovering new cocrystals via coformer-network analysisThe use of multi-component crystals, such as salts, solvates and cocrystals, is an effective way of optimizing the physicochemical and biopharmaceutical properties of active pharmaceutical ingredients (APIs) without modifying the chemical nature of the APIs . Since most APIs are produced in the form of racemic mixtures, the formation of multi-component crystals may also lead to purification of the enantiomers . Therefore, knowledge of the solid-state landscape of an API, in terms of polymorphism and multi-component formation, is of paramount importance during the design and optimization of the final drug product.
The experimental screening of new multi-component systems, and specifically cocrystals, is a labor and time intensive job and computational tools to understand and predict new cocrystals can significantly speed up the discovery of new solid forms. In this contribution, we present a data-mining approach that exploits the vast amount of information contained in the Cambridge Structural Database  in order to predict new multi-component systems. First, all information on salts, solvates and cocrystals is converted into component networks. Next, the networks are analysed to discover their organizational principles and to find the best algorithm for cocrystal prediction. These algorithms are then used to discover unknown cocrystals on the basis of the coformer network (Figure). The prediction results from this new network approach were validated for both a common coformer and an API, resulting in the discovery of several new cocrystals.
 Berry, D.J. & Steed, J.W. (2017). Adv. Drug Deliv. Rev. 117, 3–24.
 Eddleston, M.D. et al. (2012). Chem. Commun. 48, 11340–11342.
 Groom, C.R. et al. (2016). Acta Cryst. B72, 171–179.
Keywords: cocrystals, Cambridge Structural Database, networks