Patents filed for novel small molecule SARS-CoV-2 Main Protease (Mpro) inhibitors
Vancouver, BC – January 5, 2023 – Variational AI, developer of the Enki™ generative artificial intelligence (AI) drug discovery platform to redefine the economics of drug development, today announced that the company has filed two provisional patents with the United States Patent and Trademark Office (USPTO) related to new chemical entities (NCE) successfully created using generative AI. Variational AI’s protease inhibitors are among the first NCEs created via a purely generative model and will be useful in the development of COVID-19 drugs as part of the company’s coronavirus antiviral program. Of even greater significance, this method underscores the potential of generative AI to create novel, efficacious, safe, and synthesizable molecules.
“The filing of these patents is a concrete milestone not just for Variational AI, but for the applicability of generative AI to the domain of drug discovery,” said Handol Kim, co-founder and CEO, Variational AI. “Whereas adoption of AI for drug discovery is accelerating, until now it has been based on older discriminative models. Though powerful, these models are essentially just performing faster and better virtual screening of known or existing molecules. Enki™ does not screen, but rather truly creates new molecules, similar to how other generative AI models like DALL-E and ChatGPT create novel images and text based on prompts. In our case, we use the language of chemistry as prompts to describe the molecules we want.”
While generative AI has recently been applied at scale to the creation of novel images, text, music, video, and software code, the successful application of generative AI to drug discovery is still nascent but potentially transformative. Variational AI’s platform created two novel lead candidates that each bind differently and distinctly to the SARS-CoV-2 main protease, the most promising therapeutic target implicated with COVID-19 across variants and potentially other coronaviruses. The platform is based on a generative model framework known as a variational autoencoder (VAE) and created these lead candidates after being trained on large sets of public and proprietary data including older SARS-CoV (SARS) main protease data.
The candidates demonstrated potency, selectivity, and novel binding modes on the SARS-CoV-2 main protease that show potentially better safety profiles compared to the existing standard of care. Variational AI is now focused on training Enki™ for multiple drug targets in oncology and making rapid progress in this disease area, leveraging learnings from its success in antivirals, as well as other disease areas.
“Our team of machine learning researchers has been developing and applying generative models for years, long before the recent interest,” said Jason Rolfe, co-founder and CTO, Variational AI. “Drug discovery is a massively difficult undertaking, yet we continue to make rapid progress working hand-in-hand with our drug discovery team as we move to addressing unmet medical needs in oncology and beyond.”
Funding for this work was supported in part by Canada’s Digital Technology Supercluster in collaboration with adMare BioInnovations and the University of British Columbia.
About Variational AI
Variational AI uses generative AI to create novel and diverse compounds that are optimized for multiple properties to avoid the most common causes of drug attrition and increase clinical probability of success. Variational AI works with leading biopharmaceutical partners and is developing its own internal pipeline. To learn more, visit https://variational.wpengine.com.
Jordan Bouclin and Rita Murphy