The structure of the crystal-form of a pharmaceutical compound determines its properties such as polymorphism, dissolution, elasticity, compressibility, crystallinity, and habit. These crystal properties need to be tightly controlled during formulation of a pharmaceutical compound. Computational prediction of crystal structure and its properties can be helpful for early assessment of polymorph risk, and further guide the formulation process. In this work, we present a cloud-computing solution for crystal structure prediction. We present a novel dimer-expansion approach for rapid calculation of energies and gradients of crystal structures at DFT-level, that is as accurate as that of plane-wave DFT. Due to high scalability of our methodology, the computation can be divided across thousands of CPUs, allowing us to predict the crystal structure and polymorphs within a few hours to a day. The current work forms the basis for predicting more complex thermodynamic properties of the crystal that are valuable for pharmaceutical development.

Computational crystal structure and polymorph prediction

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.