Quantum Computers Simulate Protein for First Time

For years, computational chemists faced a fundamental limitation: while classical computers could efficiently model certain aspects of protein behavior, high-accuracy quantum-mechanical treatments of entire proteins remained impractical. As molecular systems increase in size, accurate electronic structure calculations on classical computers become increasingly challenging, creating a barrier to simulating biologically relevant molecules that could transform chemical, materials science, and medical research. This computational bottleneck meant researchers could only approximate the quantum behavior of large molecules, limiting their ability to predict how proteins fold, interact, and function at the most fundamental level.

Now, a joint Cleveland Clinic-IBM research team has demonstrated what was previously impossible: using quantum computing to simulate the electronic structure of a protein. The researchers modeled the 303-atom miniprotein Trp-cage using a quantum-centric supercomputing workflow and an IBM Quantum Heron r2, marking the first time quantum computing has been applied to protein simulation. Dr. Kenneth Merz of Cleveland Clinic expressed amazement at the achievement, stating, 'I'm sort of pinching myself that we were able to do it,' highlighting how this work represents a significant step toward eventually enabling accurate modeling of large, biologically relevant molecules.

The breakthrough was made possible by a novel quantum-centric supercomputing workflow that combines quantum and high-performance classical computing. The approach relies on wave function-based embedding (EWF) to fragment Trp-cage into computationally tractable pieces called clusters, with each cluster encompassing a local region surrounding an atom and its entangled neighbors. Simple clusters at the protein's edges, where atoms have minimal entanglement, are solved efficiently using classical s, while more complex clusters near the molecular core, enmeshed in intricate webs of intermolecular interactions, are assigned to quantum computers. This division of labor allows the system to leverage the strengths of both computing paradigms.

A key algorithmic innovation driving this workflow is sample-based quantum diagonalization (SQD), which addresses the combinatorial explosion of possible electron configurations as molecules grow larger. In SQD, the quantum computer samples the vast configuration space, identifying key configurations for the classical computer to focus on, with the classical computer then using this information to find solutions. Merz described a 'eureka moment' when he first saw IBM scientists present SQD, leading his team to 'drop everything' and pursue this approach, beginning with smaller molecules before scaling up to Trp-cage.

Have been extremely promising, with the workflow performing competitively with classical approaches and approaching the accuracy of the most computationally demanding among them. The researchers modeled both the unfolded and folded states of Trp-cage, a miniprotein useful for benchmarking computational chemistry s due to its compact size and features common to larger molecules, including a hydrophobic core and hydrogen bonding. Mario Motta, co-author of the paper, noted that proving this approach works for Trp-cage represents 'a step to larger molecules,' with the team initially planning to simulate just a couple of amino acids before discovering they could scale all the way up to the full protein.

Looking forward, the researchers believe the combined EWF-SQD workflow can scale far beyond Trp-cage, though they acknowledge that as molecules get larger, the task of fragmenting them, calculating complex clusters, and stitching back together becomes more complex. The team is already exploring next steps, eyeing even larger molecules as targets, while emphasizing that advancing quantum-centric supercomputing requires continued collaboration between quantum and high-performance computing researchers. This work was enabled by access to HPC resources at Michigan State University and Cleveland Clinic, similar to other recent collaborations between IBM and HPC leaders like RIKEN.

As these s mature and scale, Merz envisions QCSC workflows supporting computational pipelines for pharmaceutical research and related fields, potentially enabling scientists to build databases of simulated molecular behavior. These databases could then train machine learning algorithms to identify molecules with desired properties, which could be synthesized and tested in real life. While this represents a long-term vision, the Trp-cage simulation demonstrates that quantum-centric supercomputing can already tackle problems previously considered impractical for quantum systems alone.

The work represents an early look at quantum-centric supercomputing in action, showing how quantum computers can load-share with classical computers in hybrid workflows to solve complex scientific problems. While significant s remain in scaling to even larger molecules and improving accuracy, this demonstration provides a concrete example of how quantum computing might eventually transform molecular simulation across multiple scientific disciplines.