Quantum Computers Simulate 12,635-Atom Protein Complex, Setting New Scale for Drug Discovery

Researchers from Cleveland Clinic, RIKEN, and IBM have successfully simulated the electronic structure of a protein-ligand complex containing 12,635 atoms using a hybrid quantum-classical computing approach.

The specific molecules modeled were Trypsin—a digestive enzyme—bound to an inhibitor, and T4-Lysozyme, an immune system protein. Both were simulated in a liquid water solution to mimic their natural biological environment.[1]

This achievement represents the largest biologically meaningful molecular simulation ever performed on quantum hardware, shattering the previous record of a 303-atom miniprotein set just four months prior.

The core challenge in computational drug discovery is accurately predicting how a small molecule, known as a ligand or drug candidate, binds to a large target protein.[3]

Classical computers struggle with this task because of "electron correlation"—the complex, highly entangled quantum mechanical interactions between electrons that dictate how molecules bind and react.

To overcome this computational bottleneck, the international team utilized a framework called quantum-centric supercomputing (QCSC), which delegates different parts of the simulation to the machines best suited for them.[2]

The workflow relies on a novel algorithmic approach known as EWF-TrimSQD (Embedding Wave Function – Trimmed Sample-based Quantum Diagonalization).[2]

Classical supercomputers—specifically RIKEN's Fugaku and the University of Tokyo's Miyabi-G—first deconstruct the massive protein-ligand complex into smaller, computable fragments.[2]

The classical systems handle the simpler, less entangled regions of the molecule. The most complex clusters, where electron correlation is strongest, are offloaded to the quantum processors.[2]

Two 156-qubit IBM Quantum Heron r2 processors executed the quantum sampling. Using up to 94 qubits, the processors ran 9,200 circuits over 100 hours, collecting 1.3 billion measurement outcomes.

A critical algorithmic breakthrough allowed the team to scale the simulation without incurring exponential computational costs. Researchers recognized that quantum mechanical correlations become negligible beyond a local sphere of 7 to 10 angstroms.[2]

By restricting the quantum calculations to these localized spheres, the team achieved a 40-fold increase in system size and a 210-fold improvement in accuracy for a key workflow step compared to previous methods.

Dr. Kenneth Merz, the study's lead author from Cleveland Clinic, noted that crossing the 12,000-atom barrier demonstrates a viable framework for applying quantum methods to scientifically relevant biological problems.[1][3]

The pharmaceutical industry closely monitors these developments because accurately computing molecular energies and movements early in the discovery process could drastically shorten drug development timelines, which currently span over a decade.[3]

However, researchers and analysts emphasize transparent uncertainty regarding immediate clinical applications. The current hybrid method does not yet outperform the best purely classical approaches for protein chemistry in terms of speed or cost-efficiency.

Furthermore, the successful simulation of a protein-ligand complex is a computational milestone, not an approved drug discovery result. The healthcare applications of quantum computing remain strictly in the preclinical and computational phases.

Despite these caveats, the trajectory of improvement is striking. Moving from a 303-atom system in a vacuum to a 12,635-atom system in an aqueous solution within months suggests that quantum-centric approaches could become competitive with classical alternatives in the near future.

As fault-tolerant quantum hardware continues to evolve, this hybrid workflow is expected to scale further, potentially unlocking predictive modeling of entire biological systems and fundamentally changing how medicines are designed.