Cleveland Clinic, IBM report protein simulations spanning up to 12,635 atoms
Cleveland Clinic, RIKEN and IBM used quantum and classical computers to simulate protein complexes spanning up to 12,635 atoms. The work builds on an earlier 303-atom Trp-cage simulation.
Scientists at Cleveland Clinic, RIKEN, and IBM have used IBM quantum computers and two of the world’s most powerful supercomputers to simulate protein complexes spanning up to 12,635 atoms. These are the largest-known simulations of biologically meaningful molecules performed with quantum hardware yet, and the results were achieved in part by an innovative algorithm that optimizes how quantum and classical computers can work together, a framework known as quantum-centric supercomputing.
Using this approach, the team captured the behavior of two biochemically relevant proteins that are roughly 40 times larger than what this same method could initially achieve just six months ago. Additionally, the accuracy of the simulations in a key step of the workflow improved by up to 210 times over this same period.
The decision to explore if quantum computers could offer value in the simulation of protein complexes was motivated by challenges faced today by researchers when studying how a drug candidate could bind to a protein. This can be one of the most difficult and expensive problems in life sciences research, and one that today’s existing computational methods have struggled to exactly solve as molecules increase in size.
In this work, classical computers deconstructed the protein-ligand complexes into computable fragments. IBM’s 156-qubit IBM Quantum Heron processors, running within the IBM quantum computers at both Cleveland Clinic in the United States and RIKEN in Japan, calculated the quantum-mechanical behavior of those pieces in tandem with Fugaku at RIKEN and Miyabi-G, operated by the University of Tokyo and the University of Tsukuba. The computation required up to 94 qubits running nearly 6,000 quantum operations within certain parts of the simulation, and results were reassembled on classical computers to obtain a complete representation of the molecule.
As published on arXiv, the jump in scale was made possible by both algorithmic innovation and access to cutting-edge computing infrastructure. The novel quantum-classical hybrid algorithm, coined EWF-TrimSQD, dramatically reduced computational overhead and accelerated the ability to directly represent the chemistry of these molecular systems on quantum hardware.
The breakthrough research builds on earlier work that modeled the 303-atom benchmark molecule called Trp-cage, the first-known full quantum-centric simulation made of 20 amino acids. In that earlier demonstration, a joint Cleveland Clinic-IBM research team modeled the 303-atom miniprotein Trp-cage using a quantum-centric supercomputing workflow and an IBM Quantum Heron r2. The researchers modeled both its unfolded and folded states.
The Trp-cage workflow relied on wave function-based embedding to fragment the molecule into computationally tractable pieces called clusters. In any given protein, some clusters can be solved efficiently using classical computational methods, while larger clusters closer to the molecular core are good problems for quantum computers to solve. Stitched back together, the results of individual cluster calculations lead to a complete solution for the electronic structure of the molecule, which describes where its electrons are and how they interact.