In 1988, the then Editor of Nature, John Maddox, began an editorial with the following statement 'One of the continuing scandals in the physical sciences is that it remains in general impossible to predict the structure of even the simplest crystalline solids from a knowledge of their chemical composition'. In response to this the CCDC played a leading role in the establishment of a series of bind tests of Crystal Structure Prediction (CSP) techniques. These tests continue, with the next evaluation, again hosted by the CCDC in late 2010. The selection of appropriate target molecules and evaluation of results requires a good deal of complex research. In addition, the CCDC is generating its own, knowledge-based methods of CSP.
Using the information on molecular conformations encapsulated in the CSD, we successfully predicted the crystal structure of a molecule with 8 torsion angles as part of the 2010 CCDC Blind Test. We are interested in combining our structural informatics view of crystal structure stability with that of the ranking of predicted structures (based on energy) returned from CSP.
We are developing advanced tools and methods to analyse and predict molecular conformations in the CSD. These methods would be of extreme relevance for choosing the right starting conformations in CSP searches. In addition, we are also planning to develop software for the generation of crystal structures using CSD-biased searches of the crystal structure conformational space. By assessing ranked predicted structures using structural informatics methods such as motif analysis, hydrogen bond interactions and Full Interaction mapping it is hoped that CSP rankings can be refined to reliably provide usable information about the polymorphic landscape.