# Cleveland Clinic Simulates Large Proteins With Quantum-Centric Supercomputing > Source: > Published: 2026-05-06 02:43:37+00:00 hpc # Cleveland Clinic Simulates Large Proteins With Quantum-Centric Supercomputing At this point, quantum computing is about making incremental steps on the pathway toward fault tolerance and commercial viability. [Error correction](https://www.nextplatform.com/compute/2026/04/17/how-hpc-and-ai-digital-twins-accelerate-quantum-error-correction/5218112) needs to be addressed and the number of qubits in a given system needs to grow dramatically, and all the while the [software ecosystem](https://www.nextplatform.com/compute/2026/03/12/four-months-into-its-comeback-zapata-stakes-its-claim-in-quantum-software/5209179) and [algorithm development](https://www.nextplatform.com/compute/2026/03/31/classiq-says-quantum-is-on-its-way-but-patience-is-needed/5213539) are both ramping up. Somewhere down the road is quantum advantage, that time when a quantum system can run a useful computation more quickly, accurately, cheaply, or efficiently than the most powerful classical supercomputer. As a note: D-Wave last year announced that a smaller version of its Advantage 2 annealing quantum computer had [gained “quantum supremacy,”](https://www.nextplatform.com/compute/2025/04/01/d-wave-pushes-back-at-critics-shows-off-aggressive-quantum-roadmap/1650724) a slightly different premise than “quantum advantage” in that it refers to a quantum system that can solve any problem – useful or not – that a classical system can’t. Some have debated D-Wave’s claim. That said, even before such milestones are reached, quantum systems are increasingly [showing their usefulness](https://www.nextplatform.com/compute/2026/03/27/demonstrating-the-scientific-usefulness-of-quantum-systems/5211728) as an emerging part of computing stacks, working as another option alongside powerful GPU- and CPU-based HPC systems in hybrid quantum-classical environments. IBM and the Cleveland Clinic, working with the RIKEN Institute in Japan, gave another example of what such [quantum-centric supercomputing](https://www.nextplatform.com/hpc/2026/03/16/ibm-unrolls-blueprint-for-quantum-classical-hpc-computing/5209400) – in IBM’s terms – can accomplish now. Scientists from all three institutions used IBM’s 156-qubit Heron r2 processors running in Big Blue’s quantum systems at both the Cleveland Clinic and at RIKEN in Japan in tandem, with two powerful classical supercomputers in Japan – the Fugaku system at RIKEN and the Myaybi-G system run by the University of Tokyo and the University of Tsukuba – to simulate a Trypsin protein (below) comprising 12,635 atoms. The results not only were the largest simulation of such molecules performed with quantum hardware, but also showed what such systems can help accomplish as part of a [hybrid compute stack](https://www.nextplatform.com/compute/2025/08/27/ibm-and-amd-tag-team-on-hybrid-classical-quantum-supercomputers/1647241) and the importance of the work on algorithms to better enable quantum systems. The Trypsin simulation that the three institutions came up with also illustrated the value fragmentation, the method of breaking down workloads into manageable parts to get worked on before being reassembled into the final result. “The way to actually perform a simulation at this scale and at this size with our approach really shows that quantum-centric supercomputing is expanding to become this useful tool in science and scientific domains, especially in areas such as biology and chemist,” Jerry Chow, IBM Fellow and chief technology officer of quantum-centric computing at IBM Research, told journalists at a media briefing. “This is really exciting, and a big part of it is that we're able to integrate cutting-edge computational resources paired with new developments in algorithms and innovation in algorithms.” Drug discovery has long been a challenge that has only been able to be done approximately by classical supercomputers, and scientists have long eyed quantum computing as the tool for accelerating work in this area. A key to drug discovery is studying how a drug candidate can bind with a protein, and simulating a protein could help with what scientists say is among the most difficult and expensive problems in the life sciences fields. It’s something that neither quantum computing nor classical supercomputers can do well on their own. In this work, which is detailed in a pre-print study, the classical systems were used to deconstruct the protein-ligand complexes – which are fundamental to biological processes – into smaller fragments. The study modeled two proteins, T4-Lysozyme and Trypsin, and using 94 qubits spanning both quantum systems, ran 9,200 circuits for more than 100 hours and collected 1.3 billion measurements. “The concept of fragmentation methods is really, really simple,” said Kenneth Merz, staff scientist in Cleveland Clinic’s Computational Life Sciences Department and the study’s lead researcher. “You take a molecule, let's just say benzene. It has six carbons and six hydrogens, so you can imagine fragmenting that up into six individual carbons and six individual hydrogens. This is the way these methods work. They fragment the problem up into smaller pieces. The beauty is, if you have a single carbon with some of its environment, this can readily fit into current generation on hardware in terms of qubit counts.” The IBM quantum systems in the Cleveland Clinic (below) and at RIKEN calculated the quantum-mechanical behavior of each of the fragments, with the results reassembled by the classical Fugaku and Miyabi-G supercomputers to create a representation of the entire molecule. Central to the effort was a novel quantum-classical algorithm, called EWF-TrimSQD, which reduced the amount of computation necessary for the work and improved the representation of the chemistry of the molecular systems on quantum hardware. The result of the work was a 40-times increase in the size of a simulation over six months. (You can read the paper describing this work [at this link on Arxiv](https://arxiv.org/pdf/2605.01138).) IBM, the Cleveland Clinic, and RIKEN have working on this for almost two years. In October 2024, they started with a methane dimer, a molecule with 10 atoms. In this process, they used traditional algorithms before embracing IBM’s subspace quantum diagonalization (QCD) algorithm. The scientists moved onto a series of larger proteins, from benzene with six of each carbon atoms and hydrogen atoms, moved onto cyclohexane, (six carbons and 12 hydrogen atoms), and, in December 2025, Trp-cage, with 303 atoms. Two months later, they simulated T4-Lysozyme and its 11,608 atoms, and last month, Trypsin with 12,635 atoms. “Long story short, we were able to calculate the total energy of this whole system up to almost 13,000 atoms, and then we are able to remove this small molecule and do the same calculation and actually get an estimate of the interaction energy,” Merz said. “This is really exciting because now we can really work on proteins that are of relevance to healthcare and life science. ... We're really working on the scale that's required in computational chemistry and biology.” The procedure can also be used in other fields, he said, from battery chemistry to metal organic frameworks. “It's really a point where quantum computers and algorithms are maturing hand in hand and we are going to see [quantum-centric supercomputing](https://www.nextplatform.com/compute/2025/08/27/ibm-and-amd-tag-team-on-hybrid-classical-quantum-supercomputers/1647241) really grow to become increasingly capable to solve these fundamental problems in science and biology, chemistry, life sciences, materials, and, really, so much more,” IBM’s Chow said. “We really see that as an architecture that brings quantum computers into a core component of the modern supercomputing stack. Everybody certainly knows about the capabilities that we've gotten with supercomputing with CPUs and certainly today with GPUs, especially with GPUs in their application to AI workloads and so forth. But now we're able to really bring quantum into that mix, comparing alongside CPUs and GPUs to solve problems that are really fundamentally challenging for ASCII computing.”