On the Road to Cocrystal Prediction: A Screening Study for the Validation of In Silico Methods

The pharmaceutical industry is increasingly exploring
cocrystals
as a solution to provide improved material properties for otherwise
intractable active pharmaceutical ingredients (APIs). Researchers
have attempted to streamline the experimental process of screening
for cocrystals by developing in silico predictive tools. These tools
use intermolecular interactions, primarily hydrogen bonding, as well
as other molecular descriptors to quickly assess the likelihood of
cocrystal formation between an API and a set of small-molecule coformers.
We have developed a web-based application using three such predictive
tools to help us prioritize experimental screening against a library
of nearly 300 individual coformers. In order to validate our predictive
algorithms, three API molecules from our compound library were screened,
experimentally and with our application, against a subset of 40 coformers.
Here, we present the design of the web-based app, the experimental
work used to validate its predictions, and the relative success of
our techniques.