Validation of a Computational Cocrystal Prediction Tool: Comparison of Virtual and Experimental Cocrystal Screening Results
datasetposted on 02.01.2014 by Tudor Grecu, Christopher A. Hunter, Eleanor J. Gardiner, James F. McCabe
Datasets usually provide raw data for analysis. This raw data often comes in spreadsheet form, but can be any collection of data, on which analysis can be performed.
A virtual cocrystal screening method based on calculated gas phase molecular electrostatic potential surfaces (MEPS) of the individual components has been validated using experimental cocrystal screens reported in the literature. The noncovalent interactions of a molecule with its environment are described by a discrete set of independent surface site interaction points (SSIPs), whose properties can be calculated from the ab initio MEPS. The stability of a crystal is estimated based on pairing SSIPs such that the sum of the pairwise interaction energies is optimized. This provides a means of calculating the relative stability of a cocrystal compared with the pure components without knowing anything about the three-dimensional structures of the crystalline states. For a set of potential crystal coformers (CCF), the difference between interaction site pairing energies of different solid forms (ΔE) provides a method for ranking CCFs based on the calculated probability of cocrystal formation. The method was applied to cocrystal screens of 18 compounds that reported both hits and misses, and in most cases, the virtual cocrystal screen reproduces experimental results well. In lists of CCFs ranked by ΔE, the experimentally observed hits were significantly enriched at the top, and this indicates that virtual screening is a promising tool for focusing experimental efforts on the most promising CCF candidates.