Crystal structure prediction molecular flexibility

Methods of clustering are implemented in most packages devoted to crystal structure prediction. For flexible molecules it is always helpful to minimise the energy of the candidate structures without any constraints, before performing clustering. This ensures that the molecular geometries in equivalent structures are as similar as possible, given the accuracy of the minimisation. 

Since many critical physical and mechanical properties of pharmaceutical compounds are in large part dependent on crystal form, accurate prediction of crystal structure would be a highly valued tool in the pharmaceutical industry. Unfortunately, to date, reliable crystal structure prediction is only feasible for rigid, low molecular weight molecules, which do not represent the size and flexibility of pharmaceutical molecules. 

In spite of these early successes, computational methods and algorithms that allow packing simulations with many degrees of freedom, including unit cell parameters and molecular flexibility, were not developed until this decade, making true ab initio crystal structure predictions possible on the basis of molecular information alone. 

A few things can be said about the overall results of the four blind tests (Table 2.2) there has been some success for rigid molecules, although the predictability of the different category 1 and 2 crystal structures is variable. Molecular flexibility is a serious obstacle for current methods of crystal structure prediction only one category 3 success was achieved in the first three blind tests and, while more success was achieved for the flexible molecule in the latest test, this was partly due to the more restricted molecular flexibility of the molecule chosen that year [19]. 

While much of the development of methods has necessarily focussed on the simplest small, rigid molecules, the focus of recent crystal structure prediction studies has noticeably shifted to more complex molecular systems. This shift in the nature of targets for structure prediction somewhat reflects the feeling that, for small rigid molecules, current methods are already useful and, while they still require further refinement, there is a pressing need to extend the applicability of what has been learnt from rigid molecules to more general, flexible molecules. 

Crystal structure prediction can give insights into molecular packing and the relevant solid-state interactions of cocrystals [46,47] but is yet too time consuming to be of practical use for flexible, multicomponent systems, or even for the screening over several coformers. 

Crystal structure prediction The field of organic crystal structure prediction remains one of the best testing grounds for intermolecular potentials. Acciuades need not be as high as that needed for spectroscopic calculations, but the effects of molecular flexibility and many-body non-additivity need to be accounted for. See Price (2008, 2009) for recent reviews of this subject. For a description of dispersion-corrected DFT methods specially parametrized for organic crystals see Neumann and Perrin (2005). For a comprehensive examination of the role of detailed distributed multipole models in this field see Day et al. (2005). 

Perlstein, J., Molecular self assemblies. 3. Quantitative predictions for the packing geometry of perylenedicarboximide translation aggregates and the effects of flexible end groups. Implications for monolayers and three-dimensional crystal structure predictions, Chertu Mater., 6, 319, 1994. 

---