Cyrus Bio announces full release of rapid and accurate protein mutational free energy tool in Cyrus Bench

Seattle, WA, January 7, 2019 – Cyrus Biotechnology, Inc., a Seattle-based firm offering access to Rosetta, the most advanced and validated software available for protein modeling and accelerated drug discovery, announced today the general release of an improved tool to predict the effect of sequence mutations on protein stability. This new software allows Cyrus customers to easily and rapidly triage thousands of potential mutations for a candidate protein molecule, in turn reducing the time and expense of experimentation. The result is more efficient discovery of stable, active, and effective Biologic drugs and other proteins. The new mutational free energy software, herewith termed “ddG II”, replaces the existing free energy ddG tool in Cyrus Bench®. All current customers have immediate access to ddG II, and new customers can access this function simply by subscribing to the Cyrus Bench® suite of tools. The result of several years of work, ddG II improves on the original ddG calculation approach in Rosetta and Bench in a number of ways, including:

• More precise control of atomic movement
• An improved Rosetta energy function
• New algorithms to optimize mutations to and from proline.

The new method has been benchmarked across a newly curated set of over 750 mutations chosen to evenly represent a broad range of different types of mutational changes.

• On this new benchmark set, ddG II provides a 20% improvement in rank ordering over the original ddG, as measured by the Predictive Index⁽¹⁾.
• It improves categorization, as evaluated by the AUC from ROC analysis⁽²⁾
• It reduces the number of egregious errors by more than 50%.

As with the original ddG method, algorithmic efficiency and the significant parallelization accessible via Bench allow hundreds of ddG II calculations to be run per hour. Details of the original version of this approach are found in Park et al. 2016⁽³⁾, and the new benchmarking and refinements will be expanded upon in a forthcoming publication focused specifically on the ddG II algorithm. The newly curated benchmark set has been critical to the development of ddG II. In order to create this new set, Cyrus scientists examined existing datasets of protein mutational free energy changes and identified two significant issues with those datasets:

1. They tend to be overweighted by a handful of mutation types, particularly mutations to Alanine.
2. Some of the data suffers sign inconsistencies.

The new dataset has been curated to address both of the above issues and then used to tune the algorithms now available as ddG II. One intuitive way to measure the accuracy of a free energy change is simply by categorization – experimentally measured values that are negative should be predicted by the algorithm as negative, positive as positive, neutral as neutral. By such a categorization measure, ddG II is significantly improved and is quantitatively better across a wide range of accuracy metrics. By improving accuracy while maintaining extremely high speed and throughput, and by retaining the simple graphical user interface for this method in Cyrus Bench®, these free energy capabilities are now more useful to a much broader set of scientists in both industry and academia. For example, it is very common to screen mutations using parallelized experimental methods that integrate approaches such as robotics or high throughput spectroscopic screening. However, – even the most efficient of these experimental methods require weeks or months to carry out. By contrast, the ddG module in Cyrus Bench® can evaluate thousands of variants in a day, thereby drastically reducing both the number of mutants that need to be tested experimentally and the time required to carry out this testing. By combining speed and accuracy, researchers can use ddG in Bench to quickly test thousands of mutations, so they can focus their experimental efforts on a small set of variants instead — saving time and money while identifying better molecules in the process. These methods do not replace all work in the lab, but they make the experimentation process much more efficient, enabling more rapid discovery of valuable molecules.

1. Pearlman DA & Charifson P (2001) J Med Chem 44 3417-3423
2. Fawcett T (2006) Pattern Recognition Letters 27 861–874
3. Park H, Bradley P, Greisen P Jr, Liu Y, Mulligan V K, Kim D E, Baker D & DiMaio F (2016) J Chem Theory and Comput 12 6201-6212

Cyrus Biotechnology Inc. Cyrus Biotechnology, Inc. is a privately-held biotechnology software company offering Cyrus Bench^®, a Software-as-a-Service (SaaS) platform for protein modeling and design capabilities to accelerate discovery of biologics and small molecules for the Biotechnology, Pharmaceutical, Chemical, Consumer Products and Synthetic Biology industries.  Cyrus Bench^® is based primarily on the Rosetta software from the laboratory of Prof. David Baker at the University of Washington, the only protein engineering software experimentally proven to design new proteins completely via software. Cyrus customers include 10 of the top 20 Global Pharmaceutical firms and is financed by leading investors in both Technology and Biotechnology, including Trinity Ventures, Orbimed, Springrock Ventures, Alexandria Venture Investments, the W Fund, and others. For more information please visit