In a recent article published in the journal Scientific Reports, the researchers proposed a hybrid quantum pipeline for real-world drug discovery.
Research: A hybrid quantum computing pipeline for real-world drug discovery. Image credit: Panchenko Vladimir/Shutterstock.com
background
Quantum computing has the potential to revolutionize several scientific fields due to its superior computing power. The field of drug discovery, which requires predictive analytics and meticulous molecular modeling, can greatly benefit from quantum computing.
Recent efforts have been made to integrate quantum computing into drug design studies, which represent progress in applying advanced computational techniques to drug discovery. However, the use of quantum computing has been primarily limited to proof-of-concept studies, which do not capture the complexities of real-world drug development.
Proposed approach
In this study, the researchers established a model hybrid quantum computing pipeline that enables quantum computers to address real-world drug discovery problems. Specifically, they overcame two major challenges in computer-aided drug design, including molecular dynamics simulation and reaction barrier calculation. The goal was to accurately simulate covalent interactions and determine the Gibbs free energy profile of prodrug activation involving covalent bond cleavage.
Preparing molecular wavefunctions on quantum devices is crucial for quantum computation of molecular properties. Hence, the proposed pipeline leverages the variational quantum eigensolver (VQE) framework on the quantum computing side, as the VQE framework is suitable for near-future quantum computers, and combines a hybrid quantum-classical computing platform to efficiently manipulate and store molecular wavefunctions.
The VQE core used a parameterized quantum circuit to determine the energy of a target molecular system. A classical optimizer was then utilized to minimize the energy expectation value until convergence. The state of the quantum circuit was a good approximation of the wave function of the target molecule, and the measured energy was the variational ground state energy via the variational principle.
The researchers employed the ddCOSMO solvation model and analytical CASCI force formulas to calculate solvation energies and molecular forces for quantum mechanics/molecular mechanics (QM/MM) simulations. The interface between the classical and quantum computing sides relied on one- and two-body reduced density matrices.
Case Studies: To demonstrate the potential of the pipeline, two case studies were performed using superconducting quantum devices. In the first case study, the Gibbs free energy profile under solvent conditions for prodrug activation involving carbon-carbon bond cleavage/carbon-carbon bond cleavage prodrug strategy was studied.
In this strategy, we investigate a novel, experimentally validated prodrug activation approach applied to β-lapachone for cancer-specific targeting. Prodrug design aims to address the limitations of active drugs in pharmacodynamics and pharmacokinetics, providing an important complement to current prodrug strategies.
Simulation of the prodrug activation process requires accurate modeling of solvation effects in the human body, which we achieve by implementing a generic pipeline enabling quantum computing based on a polarizable continuum model (PCM) of solvation energy.
In the second case study, QM/MM simulations were used to investigate a covalent inhibitor of Kirsten rat sarcoma viral oncogene (KRAS) (G12C), where the energy changes were closely monitored and the time costs were compared based on quantum and classical computers.
KRAS is a common protein target in several types of cancer, where it plays a key role in the RAS/mitogen-activated protein kinase (MAPK) signaling pathway and has a profound effect on cell survival, growth, and differentiation. Mutations in the KRAS protein, especially the G12C variant, are common in several cancers, including pancreatic and lung cancers.
Sotorasib, a covalent inhibitor that targets this protein mutation, has shown the potential to provide a more specific and long-lasting interaction with the KRAS protein. Quantum computing can improve the understanding of these target-drug interactions through QM/MM simulations, as such interactions are of great importance in the computational validation stage after drug design.
Significance of the study
Based on the results of two study cases, the researchers demonstrated the potential of the proposed hybrid quantum computing pipeline to solve real-world drug design problems. The study showed the plug-and-play advantages and versatility of the pipeline.
Furthermore, the effectiveness of quantum computing in a scenario involving Gibbs free energy profile calculation/profiling of covalent bond cleavage for prodrug activation was demonstrated. The time required for energy expectation value calculation was 40 seconds. Furthermore, the calculation of the one-body reduced density matrix in the active space included three additional expectation value measurements. Therefore, the overall quantum computing kernel time cost was about 60 seconds, which was consistent with the experimental results.
Strong and specific binding was observed between the targeted mutations and sotorasib, providing important insight into the potential efficacy of the drug. This understanding is important for future rational inhibitor design targeting similar mutations. In summary, this work demonstrated the feasibility of integrating quantum computing pipelines into practical drug design workflows.
Journal Reference
Li, W.et al. (2024). A hybrid quantum computing pipeline for real-world drug discovery. Scientific Reports, 14(1), 1-15. DOI: 10.1038/s41598-024-67897-8, https://www.nature.com/articles/s41598-024-67897-8
