Crystal Structure Prediction

The crystal form of a drug plays an important role in its solid-state formulation. Finding novel and more stable crystal forms of a drug molecule, after the drug has been approved, can pose significant risks for pharmaceutical companies.

As all stable crystal forms of a drug cannot be obtained readily through experiment, computational predictions can assist with identifying the potential risk of polymorph, as well as finding crystal forms that are better suited for formulation with regards to physico-chemical properties. 

Crystal structure prediction (CSP) of drug-like molecules poses a variety of challenges, including unknown tautomer state, novel chemical motifs, high degree of flexibility, intricate hydrogen bonding networks, and others.

To meet the challenges of CSP, OpenEye offers a custom CSP solution that is built on our cloud-native platform Orion.™  At each stage of the study, we have developed novel methods for parallelizing the problem, enabling us to exploit the massive scaling afforded by Amazon Web Services. Furthermore, we can apply a variety of force-field and quantum mechanics (QM) energy models within the study for varying levels of accuracy and scope. 

Traditional approaches to CSP of a drug molecule can take weeks to months. A major bottleneck in these approaches is the use of periodic QM lattice calculations (e.g. plane-wave density-functional theory), which are not very parallelizable and scale poorly.

OpenEye has developed a novel approach for the optimization of crystal structures using QM energy models that is highly parallelizable, and scalable to 100,000 processors. Utilizing this approach, we are able to perform CSP on drug-like molecules within the wall clock time of a few days, enabling the exploration of a variety of ideas for optimizing drug molecule formulation within a reasonable amount of time. Assessing the polymorph risk of a drug early in the drug development process can save time and effort down the pipeline. 

In an effort to evaluate our methodology, we performed several blind challenges in collaboration with GSK on drug-like molecules, resulting in the successful advancement of several novel molecules to various stages of clinical trials.­­

To find out how a collaboration with OpenEye and Orion could help with your CSP evaluations, contact us at


Ab initio prediction of small molecule crystal structures in Orion™


  • Perform CSP on drug-like molecules within a few days – instead of months – using the cloud-native platform Orion™

  • Apply various force-fields and QM energy models for a broader range of results

  • Assess the polymorph risk of a drug early in the development process to save time and effort down the pipeline