Scientists’ Ventures: Harnessing AI and Quantum Computing to Create Novel Peptides

Researchers have effectively demonstrated that a quantum computer can enhance the precision and scope of generative artificial intelligence models used in drug discovery. Remarkably, this was accomplished using their free time and leftover funds from other endeavors.

The team from the Technical University of Denmark utilized their generative AI model for protein prediction alongside a compact quantum computer created by British startup ORCA Computing. This collaboration improved AI performance by integrating quantum systems with traditional processors. The researchers employed this hybrid method to create novel peptides—short amino acid chains—designed to bind to specific proteins in the body, which is a vital component of vaccine development.

The research team dedicated their weekends and combined unspent project funds because “most innovative science is too intimidating for foundations,” stated DTU professor Timothy Patrick Jenkins, who led this initiative.

Laboratory synthesis and testing of the peptides revealed that the model produced a higher success rate in peptide creation compared to classical methods, particularly where training data was scarce.

The researchers believe this technology could expedite the creation of personalized immunotherapies and vaccines, as well as enhance drug efficacy in underrepresented populations.

“We needed to demonstrate real-world connections to convince skeptics of our predictions,” Patrick Jenkins remarked to WIRED. Quantum computing is still an emerging field and is under significant scrutiny due to the complexities involved in constructing these machines and effectively applying them to resolve issues.

Even Patrick Jenkins initially hesitated to delve into this technology: “I was a huge quantum skeptic,” he chuckled, thinking any applications to his work would be “decades away.”

His team leverages big data and AI to identify proteins that could lead to new, cheaper immunotherapies, often funded by the Novo Nordisk Foundation. While many biological model creators crave more data, his team faces a notable challenge due to the scarcity of comprehensive genetic information from diverse populations, as most medical research has concentrated on Western demographics. This can complicate the development of peptides effective for understudied groups like those in Asia and Africa.

The team theorized that integrating a quantum computer into their workflow might produce a more diverse array of peptides, particularly for targets with limited data, given that these machines showed a similar effect in image generation.

While this newly developed process isn’t set to revolutionize research immediately—quantum computers currently lack the capacity to handle full-scale, advanced AI models—better results can still be attained using classical computing.

“Quantum still isn’t very powerful, so the complexity we were able to encode wasn’t a typical-sized antibody, which is our usual focus,” explained DTU PhD student Jonathan Funk. Additionally, discovering a peptide that binds to a specific gene is merely one step in the vaccine development process and won’t independently result in successful drugs.

“It’s no surprise that many industry leaders view quantum as ambiguous and distant,” ORCA Computing CEO Richard Murray told WIRED, partly because the technology “simply hasn’t offered clear near-term examples of utility.”

He noted that this study is significant as it illustrates a near-term commercial application for quantum technology. His company is also exploring its potential through partnerships with oil giant BP in chemistry and automaker Toyota to enhance design efficiency.

The DTU team aims to assess whether this workflow can be applied to more cutting-edge models and larger proteins. “We needed this as a straightforward way to validate that we actually have the opportunity to make substantial progress,” said Patrick Jenkins, emphasizing that generative AI workflows are especially valuable in neglected diseases that receive little funding. He is also considering using a quantum computer to refine his generative AI techniques for crafting synthetic antidotes for snakebite venom.