Quantum Computing, AI Target Undruggable Cancer ProteinQuantum Computing, AI Target Undruggable Cancer Protein

Researchers from the University of Toronto and Insilico Medicine have developed an approach combining quantum computing and AI to design novel cancer drugs.  

Their study targeted the KRAS protein, a key driver in 25% of human cancers including up to 90% of pancreatic cancers. KRAS mutations are deemed undruggable due to the protein’s smooth surface that lacks suitable binding pockets.

The team used a hybrid quantum-classical model to generate potential inhibitors for KRAS, training it on a dataset of 1.1 million molecules. This included 650 experimentally validated KRAS inhibitors and 250,000 molecules sourced from the open-source virtual screening platform VirtualFlow.

The researchers used Insilico Medicine's generative AI engine, Chemistry42, to screen the molecules and identify the 15 most promising candidates for laboratory testing. Of these, two molecules demonstrated a strong ability to target multiple mutated forms of KRAS in live cells, demonstrating their potential as anti-cancer drugs.

According to the team, the study marks the first instance of using a quantum-generative model to yield experimentally confirmed biological hits, indicating the practical potential of quantum-assisted drug discovery in producing viable therapeutics.

Related:Quantum Industry Weighs in on Nvidia CEO’s ‘Decades Away’ Remarks

The findings also showed that the effectiveness of the model learning correlated with the number of qubits used, suggesting the scalability potential of quantum computing resources.

“It’s an exciting time to be working at the interface of chemistry, quantum computing and AI,” said project director Alán Aspuru-Guzik.

“This first-of-its-kind study shows that AI, with the help of quantum computers, can successfully find molecules that interact with biological targets.”

While the results are promising, the researchers acknowledge that this is a proof-of-principle study and does not yet demonstrate a significant advantage over classical methods.

However, as quantum computers continue to advance, the integration of quantum computing into AI-driven drug discovery pipelines could accelerate the development of new therapies, particularly for targets previously considered undruggable.

The team now plans to apply its hybrid quantum-classical model to other challenging protein targets and optimize the design of the two lead compounds against KRAS and carry out further pre-clinical testing on these compounds.