IBM–Moderna push mRNA design past 42 to 60 nucleotides

IBM and Moderna have set a new mark in quantum-assisted bioinformatics, modeling the secondary structure of a 60-nucleotide mRNA sequence on IBM’s R2 Heron processor. The result surpasses the prior quantum record of 42 nucleotides and targets a persistent bottleneck in vaccine design.

The team combined 80 qubits with a conditional value at risk variational approach to navigate an enormous space of base-pairing possibilities. By optimizing for lower free-energy configurations, the method expands the range of candidate structures that can be explored quickly and systematically.

Why this milestone matters

Predicting mRNA secondary structure underpins stability, translation efficiency, and manufacturability. Traditional algorithms struggle when structures include complex motifs like pseudoknots. These features are biologically important but computationally demanding, forcing many classical pipelines to exclude them.

Quantum techniques can reframe the problem as an energy-minimization task across a vast combinatorial landscape. The approach broadens the scope of search and reveals a variety of low-energy folds that classical heuristics might overlook.

Did you know?
Some mRNA structures include pseudoknots, entangled folds that are vital for biological function yet notoriously hard to predict with classical algorithms.

Inside the quantum method

The researchers mapped base-pairing possibilities onto qubits and used a CVaR-based variational routine to prioritize promising solutions. CVaR, adapted from finance, emphasizes the best-performing tail of candidate states, accelerating convergence on stable folds.

Running on R2 Heron with 80 qubits, the team balanced circuit depth and noise while preserving structural signal. Careful ansatz design and error-mitigation strategies helped stabilize readouts across repeated evaluations.

Beyond the 60-nt record

In simulation, tensor-network tools extended the same strategy to much larger problem sizes. These noiseless tests suggest headroom to scale toward longer sequences as hardware fidelity and compilation improve.

Forthcoming work targets deeper circuits and more qubits to capture additional non-local interactions. The roadmap anticipates higher gate counts while maintaining tractable runtimes for variational optimization.

ALSO READ | First in 35 years: Patient’s own insulin returns

Implications for vaccines and therapeutics

Folding shapes influence mRNA durability and expression levels, which in turn affect dose, immune response, and shelf stability. A broader search over candidate structures can accelerate lab validation and streamline design cycles.

Rather than replacing classical tools, quantum routines slot into specific bottlenecks like complex folding landscapes. The goal is a hybrid discovery loop that delivers more viable designs to experimental teams faster.

Building a quantum-enabled pipeline

Moderna’s approach complements existing modeling with targeted quantum steps where combinatorics explode. Classical engines still handle large-scale screening, data integration, and downstream analytics efficiently.

The integrated pipeline emphasizes verification and benchmarking against state-of-the-art classical predictors. Practical value comes from diversity in solutions and measurable gains in experimental success rates.

What to watch next

Key milestones include error-rate reductions, deeper variational circuits, and calibration techniques that preserve delicate structural signals. Standardized benchmarks against leading classical methods will help quantify real-world advantages.

As platforms mature, expect joint advances in compilers, error mitigation, and problem encodings tailored to RNA biology. The forward focus is clear: translate quantum modeling gains into mRNA constructs with superior performance in the lab and clinic.