Insider Brief
- Researchers used a 36-qubit trapped-ion quantum computer and a specialized algorithm to solve protein folding problems involving up to 12 amino acids, reportedly the largest such demonstration to date on real quantum hardware.
- The system applied a non-variational quantum optimization method, BF-DCQO, to find optimal or near-optimal folding configurations for three peptides, as well as solutions to MAX 4-SAT and spin-glass problems.
- The study highlights the advantages of all-to-all qubit connectivity and circuit pruning in managing complex, higher-order optimization problems on current-generation quantum devices.
Scientists are interested in understanding the mechanics of protein folding because a protein’s shape determines its biological function, and misfolding can lead to diseases like Alzheimer’s and Parkinson’s. If researchers can better understand and predict folding, that could significantly improve drug development and boost the ability to tackle complex disorders at the molecular level.
However, protein folding is an incredibly complicated phenomenon, requiring calculations that are too complex for classical computers to practically solve, although progress, particularly through new artificial intelligence techniques, is being made. The trickiness of protein folding, however, makes it an interesting use case for quantum computing.
Now, a team of researchers has used a 36-qubit trapped-ion quantum computer running a relatively new — and promising — quantum algorithm to solve protein folding problems involving up to 12 amino acids, marking — potentially — the largest such demonstration to date on real quantum hardware and highlighting the platform’s promise for tackling complex biological computations.
In a study published on the pre-print server arXiv, researchers from Kipu Quantum and IonQ reported that their system, running a specialized algorithm called BF-DCQO, consistently found optimal or near-optimal folding configurations for three biologically relevant peptides. The same setup also solved large Boolean satisfiability and spin-glass optimization problems, all of which are known to be computationally hard for classical systems.
Solving a Central Problem in Biochemistry
Protein folding is the process by which chains of amino acids assume their functional three-dimensional shape, and, as mentioned, is crucial for biological activity.
In this study, researchers mapped the folding process onto a lattice and expressed it as a higher-order binary optimization (HUBO) problem, a type of problem characterized by complex energy functions that are difficult to minimize. Think of a HUBO problem as trying to find the lowest point in a vast, jagged mountain range where every peak and valley depends on multiple, interlinked decisions made all at once. In this analogy, the lowest point represents the most stable protein fold — the solution nature is trying to find.
These HUBO problems were then solved using a new quantum algorithm — Bias-Field Digitized Counterdiabatic Quantum Optimization (BF-DCQO) — running on IonQ’s fully connected trapped-ion hardware.
By encoding protein folding into a quantum mechanical framework, the team effectively reframed it as a ground-state search problem. The energy of different folding configurations was represented by the energy of a quantum system’s Hamiltonian, which is a mathematical function that defines how the system evolves.
Quantum Protein Solver Hits New Size Milestone
The folding problems studied involved three peptides of 10 to 12 amino acids. To simulate folding, each turn in the protein chain was encoded using two qubits, and interactions between non-consecutive amino acids were modeled using known contact energies. The approach translated into quantum circuits requiring up to 33 qubits and over a thousand interaction terms.
Results showed that the BF-DCQO algorithm, coupled with minimal post-processing, consistently returned the correct lowest-energy structures. The method was robust to hardware noise, especially when the researchers used circuit pruning techniques to reduce quantum gate counts.
The proteins selected — chignolin, a synthetic β-hairpin; a head activator neuropeptide; and a segment of the immunoglobulin kappa joining gene — are all important in biochemistry or neuroscience.
The investigation represents a milestone for trapped-ion quantum computers, according to the authors of the study.
The team writes: “Notably, these instances represent the largest protein folding instances successfully solved on trapped-ion hardware to date, highlighting the value of combining our adapted protein folding solver based on BF-DCQO with trapped-ion hardware.”
Testing Quantum Optimization Beyond Biology
The algorithm’s utility extended beyond biology. The team also applied it to two other categories of hard optimization problems: MAX 4-SAT and spin-glasses.
MAX 4-SAT is a logic problem widely used in computing, where the goal is to find the assignment of variables that satisfies the most logical clauses. In other words, it’s like a logic puzzle where each rule depends on four yes-or-no decisions, and the goal is to choose the answers that satisfy as many rules as possible. Computer scientists rely on this frequently because it models real-world constraint problems, from scheduling to circuit design.
The researchers tested instances involving up to 36 variables and thousands of constraints. Their trapped-ion system, leveraging the BF-DCQO algorithm, found the correct solutions in all but a few of the cases — and those were only off by one or two satisfied clauses.
The third problem class, spin-glasses, represents disordered magnetic systems where every spin interacts with every other. These systems are a classic testbed for optimization algorithms because they feature an especially rugged energy landscape. The study tested 36-variable spin-glass instances, and in two of three cases, the quantum processor identified the exact ground state.
It’s a task similar to the one spotlighted in D‑Wave’s recent quantum advantage experiment, which sought to demonstrate a form of quantum advantage on a real-world spin-glass problem, though subsequent studies continue to add scientific debate about the scope of the result.
Algorithm and Hardware Synergy
The BF-DCQO algorithm avoids the limitations of variational quantum algorithms, which often struggle with scalability and “barren plateaus” where optimization gradients vanish. Instead, BF-DCQO dynamically updates bias fields to steer the quantum system toward lower energy states with each iteration. It draws from principles in adiabatic evolution and counterdiabatic control but is designed to be digital and compatible with current noisy quantum hardware.
To manage hardware limitations, such as gate fidelity and circuit depth, the team introduced pruning methods that cut small-angle gate operations from the circuits. These reductions proved essential: without pruning, the number of quantum gates would have exceeded what today’s hardware can support.
The researchers explained that IonQ’s hardware — based on trapped ytterbium ions — offers all-to-all connectivity, meaning any qubit can interact with any other. This capability made it particularly suitable for solving dense problems like protein folding and spin-glass models, which involve interactions across many components.
Limitations and Next Steps
While the results were encouraging, the study acknowledges several limitations. The folding models used were lattice-based and didn’t account for full molecular dynamics or chemical environments. Additionally, the post-processing step — which involves a classical algorithm to refine near-optimal quantum results — was critical in reaching optimal solutions in several cases.
“When optimal solutions are not directly obtained, we additionally apply a greedy local search algorithm to get better solutions by mitigating potential bit-flip and measurement errors, which highlights the favorable interplay between classical and quantum algorithms,” the team writes.
Future directions might include improving quantum circuit compilation, scaling the approach to longer protein chains and integrating more realistic folding models. The authors suggest that the synergy between quantum hardware and problem-specific quantum algorithms like BF-DCQO could lead to practical quantum advantage in molecular optimization and other HUBO problems.
The paper on arXiv dives in deeper technologically than this summary story, so reviewing the study for more exact technological detail is recommended. ArXiv is a pre-print server, meaning the work has not officially been peer-review, a key step of the scientific method.
The researchers include: Sebastián V. Romero, Alejandro Gomez Cadavid, Pavle Nikačević, Enrique Solano, and Narendra N. Hegade, all from Kipu Quantum. Romero and Cadavid are also affiliated with the University of the Basque Country UPV/EHU. The IonQ Inc. team members include Miguel Angel Lopez-Ruiz, Claudio Girotto, Masako Yamada, Panagiotis Kl. Barkoutsos, Ananth Kaushik and Martin Roetteler.
0 Comments