The Challenge

The SARS-CoV-2 virus binds to human cells using its spike protein, which interacts with the ACE2 receptor. Vaccines work by targeting this binding mechanism. However, the spike protein mutates rapidly, making it difficult to predict and respond to changes using traditional lab methods. Characterising these interactions experimentally is slow and expensive. The challenge was to use computational approaches to predict mutations of concern before they appear. This required simulating the spike-ACE2 binding at quantum-level accuracy, a task that demands extreme computational power and novel algorithms.

The Background

The ACE2 receptor is the entry point for SARS-CoV-2, and its interaction with the virus’s spike protein determines infection potential. The virus evolves quickly, with mutations altering binding affinity and potentially evading immune responses. Quantum mechanical (ab initio) simulations are the most accurate way to study such interactions. However, applying them to full protein systems is typically infeasible due to their computational cost. With the advent of exascale computing and advances in algorithmic efficiency, it became possible to perform full quantum mechanical modelling of large biological systems like spike–ACE2.

The Solution

The BigDFT software, optimised by the MaX Centre of Excellence for exascale platforms, was used to simulate the spike–ACE2 system. Leveraging the Quantum Mechanics Complexity Reduction (QM-CR) method, the team performed electronic structure calculations for systems of ~13,000 atoms. Four spike protein variants were analysed, including the original Wuhan strain, Omicron, and two Omicron-based mutations. By evaluating the energy contributions of individual amino acids, researchers identified which mutations would enhance binding. This predictive model provided a real-time tool for anticipating viral evolution and guiding experimental efforts.

The Achievement

The main achievement was the accurate and early prediction of the A484K mutation, which enhances the binding affinity of the spike protein to ACE2. At the time of simulation in December 2021, this mutation had not been observed in any variant. Later experiments validated the prediction, and by August 2023, A484K was detected in variant BA.2.86. This milestone demonstrates that quantum mechanical simulations, powered by HPC, can anticipate real-world viral changes with precision and timeliness—offering a significant advantage for pandemic response strategies.

The Impact

This work represents a breakthrough in how we understand and respond to viral threats. Predicting likely mutations before they spread allows health systems to proactively design vaccines and therapeutics, shortening development timelines. For pharmaceutical and biotech industries, such predictive tools could revolutionise target discovery and validation, while governments and international health agencies gain a powerful tool for epidemic forecasting and preparedness. The societal benefit is immense: faster response to emerging threats can save lives and reduce economic disruption caused by pandemics.

Key Takeaways

This success proves the potential of HPC-powered quantum simulations in accelerating biomedical discovery, strengthening pandemic readiness, and supporting rapid response frameworks:

• A484K mutation predicted 20 months before epidemiological appearance.
  
• Real-time simulation of ~13,000 atoms completed within a few hours.
  
• Demonstrated order-of-magnitude efficiency over traditional electronic structure methods.
  
• Accurate, residue-level analysis of spike–ACE2 binding mechanisms.
• Established computational pipeline for future pandemic risk assessment