The Cambridge Crystallographic Data Centre (CCDC) has announced the launch of CSD-Theory software for use in small molecule pharmaceutical development. This latest addition to the CCDC’s software toolbox delivers easy storage, management, search and display of crystal structure prediction (CSP) results alongside experimental structures on a web-based platform and an API interface. The new software, developed in collaboration with six major pharmaceutical companies, will enable organisations to incorporate and share CSP knowledge earlier and more easily in small-molecule drug development—reducing risks and shortening the time to market.
Small molecule therapeutics are regularly administered in crystalline, solid forms which undergo extensive testing to ensure stability, efficacy and safety. However, the 3D structure and packing of the crystalline form significantly impacts the properties of the therapeutic and, to further complicate matters, it is possible for a small molecule to have multiple crystal forms (known as polymorphs). These polymorphs have the same molecular structure, but different packing. Predicting alternative polymorphs is both very important and very complex; predictions can produce hundreds, if not thousands, of possible crystal structures. Varying polymorphs can be unstable in storage, ineffective, or at worst unsafe, which can, and has led to, drugs being pulled from market—costing millions. For example, rotigotine, the active ingredient in transdermal patches for Parkinson’s disease, was withdrawn for reformulation due to an unexpected polymorph in 2007.
The ability to reliably predict the occurrence of polymorphs in silico would enable pharmaceutical companies to identify and mitigate such risks earlier, as well as identify and progress suitable candidates faster with less issues. While CSP is already a key part of the drug development process for many companies, the science and tools are still evolving. Exploration of CSP data is a bottleneck, with data often confined to a handful of computational specialists.
As a registered charity, an active centre of research and a producer of leading informatics software, the work that the CCDC does in collaboration with academic institutions, pharmaceutical companies and external software providers is critical to advancing methods in structural science, developing new approaches and expanding the utility of existing tools.
The new CSD-Theory software provides a user-friendly, interactive web interface that simplifies the CSP data storage, management, search and display, enabling users of all levels to access and share critical information across teams. The additional API component allows CSP data to be stored ready for AI or ML analysis. Capable of handling the full spectrum of CSP-data types and providing a platform to easily share results, CSD-Theory is set to drive forward the data-first approach in pharmaceutical drug development.
As well as the new software, the CCDC has an extensive and growing suite of tools to aid molecule discovery and materials design—including the Cambridge Structural Database (CSD)—the world’s repository for small-molecule organic and metal-organic crystal structures. The CCDC also runs collaborative initiatives to drive advances in CSP such as the CSP Consortium and CSP Blind Test.
Similar to the CASP protein structure predicting challenge, which saw tech giants such as Google push the boundaries of protein folding prediction in 2020, the CSP Blind Tests enable researchers to test novel methods for small-molecule CSP in a controlled, unbiased setting—facilitating the development of computational techniques to predict crystal structures. The chemical systems tested are unpublished crystal structures sourced by the CCDC, overseen by an external referee from crystallographers across academia and industry. Previous years have demonstrated that these Blind Tests translate to real world results that have the potential to revolutionize the pharmaceutical industry.
The current, 7th CSP Blind Test began in October 2020 and participants have 18 months to make their predictions. This year has already seen a record number of participants—42 people in 18 countries—making it the most popular test to date. This demonstrates the growing interest and need for fast and accurate CSP and the software tools such as CSD-Theory to effectively manage, use and share data, that in turn enables pharmaceutical companies to incorporate this critical information into small molecule drug development.