Quantum computing and artificial intelligence (AI) are ushering in a new era of drug discovery, as demonstrated by groundbreaking research led by Insilico Medicine and the University of Toronto. Their recent study, published in Nature Biotechnology, highlights how these advanced computational methods can identify molecules targeting the notorious KRAS protein—once deemed 'undruggable' due to its complexity.
KRAS mutations are present in approximately one out of every four cancers, driving uncontrolled cell growth and posing significant challenges for researchers. Despite their prevalence, only two FDA-approved drugs currently exist for specific KRAS mutations, extending patient lifespans by merely months compared to traditional chemotherapy. This scarcity underlines the urgent quest for more effective therapies.
Alán Aspuru-Guzik, project director and professor at the University of Toronto, expressed enthusiasm for the intersection of chemistry, quantum computing, and AI. "It’s an exciting time to be working at the interface of chemistry, quantum computing and AI," he said, emphasizing the research's innovative approach. The main objective was to design novel compounds targeting KRAS, utilizing the power of quantum and generative AI alongside classical computing methods.
The collaborative team employed a hybrid quantum-classical model, training it on over 1.1 million molecules. This dataset included validated KRAS inhibitors and analogs from the ultra-large virtual screening platform, VirtualFlow. By leveraging Insilico’s generative chemistry platform, Chemistry42, they screened potential candidates and identified 15 molecules worthy of laboratory testing.
Out of these, two candidates emerged with strong potential to target multiple KRAS mutations effectively. Their innovative approach could drastically cut down the preclinical phase of drug discovery, which traditionally spans several years. "With computational approaches like this, we have the potential to shorten the preclinical phase of drug discovery by years," Igor Stagljar, co-investigator and professor at the Donnelly Centre of U of T, stated.
While the findings of this study indicate the potential of quantum computing to accelerate early-stage drug discovery, it stops short of declaring superiority over traditional methods. Aspuru-Guzik noted, "Even though we show quantum computing can help with drug discovery, it doesn’t mean it's necessarily the best at it." Instead, their research serves as proof-of-principle, emphasizing the promise of integrating quantum technologies within drug development pipelines.
Alex Zhavoronkov, CEO of Insilico Medicine and one of the study’s authors, remarked on the significance of this work. "Success with KRAS could set a precedent for addressing other challenging protein targets," he said, highlighting the utility of combining quantum computing with classical drug discovery methods.
The technology demonstrates its potential not only against KRAS but also against other proteins long considered difficult to target, such as the ionotropic glutamate receptor, the SARS-CoV-2 main protease, and the β2-adrenoceptor. Zhavoronkov stressed, "While these targets were not synthesized or experimentally validated in this study, the quantum-classical approach achieved promising docking scores and showcased potential for generating viable molecular candidates."
Researchers first established the hybrid model using millions of molecules, showing its capacity to yield diverse and synthesizable leads. The two promising KRAS inhibitors, ISM061-018-2 and ISM061-022, demonstrated specific activity against various mutant KRAS forms without significant toxicity, reinforcing the hybrid approach's validity.
Ghazi Vakili, the study’s first author and postdoctoral fellow, expressed optimism for the future. "Our current study highlighted quantum computing’s ability to accelerate early-stage drug discovery, particularly in generating diverse and synthesizable leads.", showcasing how effectively this technology can optimize drug discovery methods.
Discussions are already underway to leverage the hybrid model for developing new candidates against other previously deemed undruggable targets, through the principles outlined during their work on KRAS. Gazhi Vakili remarked, "We plan to publish additional findings showcasing the expanded capabilities of our approach when applied to more targets." Specifics about the development of KRAS-targeting therapies remain pending, but plans to expedite and broaden the scope of research are clearly outlined.
The collaboration, powered by the Acceleration Consortium, seeks to unite academic, industry, and governmental efforts to hasten the discovery of breakthrough therapies. Supporting organizations include the Canada 150 Research Chairs program, Genome Canada, and the Defense Advanced Research Projects Agency.
Alex Zhavoronkov emphasized the gravity of their work, referencing data indicating up to 85 percent of human proteins are presumed 'undruggable.' This reality presents a substantial hurdle to developing newer cancer treatments, and it is here where the capabilities of AI stand to make great strides.
At the forefront of pushing these boundaries, the research team effectively integrates both quantum computing and AI technologies, potentially transforming the future of drug discovery. With the world of cancer treatments poised for evolution, this innovative research paves the way for tackling unmet medical needs.