After years of waiting, expectations are high that quantum computing will soon start making its way into practical applications. A good exemplar of progress is an ongoing collaboration between IBM and Moderna in drug discovery that blends quantum computing and HPC.
Moderna is the mRNA specialist best-known for its mRNA-based COVID vaccines. A key challenge for Moderna is developing the mRNA technology instructions that will accurately instruct the body on how to make the proteins that can treat diseases. For any given protein, there’s a huge number of possible mRNA sequences that could encode it, making optimization a complex task. (Link to Moderna mRNA primer)
IBM and Moderna have been exploring the use of variational quantum algorithms to help predict mRNA-based molecular structure and dynamics. Today, IBM posted an article profiling the POC work which has already yielded significant results.
“Our goal is to improve human health,” said Alexey Galda, Associate Scientific Quantum Algorithms and Applications, Moderna. “We believe it’s critical to explore every available tool — including quantum computing — to scale our progress today rather than wait for the technology to fully mature in the future.”
The Moderna and IBM researchers used Conditional Value at Risk (CVaR), a risk-assessment technique used in finance, to improve the performance of VAs and to find optimal solutions to complex optimization problems. CVaR helps investors assess the tail risk of a portfolio, to estimate the possible loss of investment in worst-case scenarios. In quantum computing, CVaR focuses the optimization process on the lower tail of the energy distribution, effectively targeting the most promising solutions.
CVaR mitigates variance by focusing the optimization on the lowest-energy portion of the measurement distribution, effectively steering the classical optimizer toward more promising solutions while reducing sensitivity to noisy outliers. Because CVaR operates as a lightweight classical post-processing step, it can enhance VAs without adding significant computational overhead.
It is, in essence, a hybrid approach leveraging quantum and HPC.
As reported by IBM, the Moderna-IBM team applied CVAR-based VAs to the mRNA secondary structure prediction problem. The result was one of the largest and most advanced VA executions ever realized on quantum hardware-and a demonstration of the real potential of quantum computing in aiding in Moderna’s research.
“In 2024, this work reached a record-setting scale for a quantum secondary structure simulation, involving up to 80 qubits and mRNA sequence lengths up to 60 nucleotides.
To the best of the authors’ knowledge, no one had ever simulated sequences of even 42 nucleotides on a quantum computer,” reports IBM.
“In work that will be published later in 2025, the researchers applied the same methodology to problem sizes of up to 156 qubits involving 950 non-local gates, a measure of circuit complexity. They also presented a new approach to this sort of problem, called instantaneous quantum polynomial (IQP) circuit-based quantum optimization. This sampling-based approach, similar to CVaR-based VAs, allows for the most efficient use of quantum and classical resources in a joint quantum high-performance computing (HPC) environment.”
Moderna says its end goal is not to replace classical computing with quantum methods but to build a near-term quantum-enabled biotechnology pipeline.
“Very often people only think of quantum outperforming classical. That’s not necessarily the goal. It’s also valuable if your quantum tool can offer you a more diverse set of solutions-a more diverse set of molecules to generate and go test in the wet lab,” Galda said. “Having this additional tool with its own very specific qualities is extremely valuable for the computational problems that are core bottlenecks in our workflow. I think the most realistic scenario is that quantum will augment our classical computation and offer certain advantages in certain areas.”
Jay Gambetta, Vice President of IBM Quantum, has been saying for some time that he expects that the world will see the first examples of quantum advantage by 2026. One route to quantum advantage, he notes, is refining and improving heuristic methods which is what Moderna and IBM have demonstrated.
It’s best to read the full article.
Link to IBM article, https://www.ibm.com/case-studies/moderna
Source: IBM