In helping to predict the physical structures of organic molecules, the National Grid Service has aided medical breakthroughs, writes Judy Redfearn
Sally Price, professor of theoretical chemistry at University College London, is taking advantage of the National Grid Service's ability to store and share data. She leads a project to develop a computational method for predicting the many crystal structures that an organic molecule can adopt. The NGS hosts the results of these predictions in the Computed Crystal Structure (CCS) database. "Using the NGS has enabled us to store the many hundreds of data files and allows easy access for our users," Price says.
Being able to predict crystal structure is an important goal for the chemical industry because the action of an organic molecule depends not only on its chemical composition but also on its physical shape. For example, the crystal structures in cocoa butter can affect the taste of chocolate; and crystal structure can determine a drug's efficacy by affecting its solubility and hence its rate of absorption into the bloodstream.
Predicting crystal structure is computationally demanding. It involves identifying the most stable structures for each molecule studied out of numerous possibilities. Work at UCL resulted in a method that uses a variety of programs to estimate the properties of all these possibilities; each structure produces more than a dozen files. The files are stored in the CCS database together with metadata (data about the data) that detail how each study was done. The method has had some notable successes, most recently with the prediction of a new solid form of progesterone, a hormone used in oral contraceptives and hormone replacement therapy. Subsequent experiments succeeded in making the predicted form, which is very stable.
Earlier successes included the prediction of previously unknown crystal structures for aspirin and the Alzheimer's drug piracetam, which were later (or independently) found experimentally.
Users can gain access to the CCS database via a data portal and web interface. First, however, they need an NGS grid certificate and a security code. Security is tight to avoid fraudulent claims that predicted structures exist before experimental evidence is available.