Brief Description

The emergence of mRNA-based therapeutics has ignited a paradigm shift in the landscape of medicine. Accurate prediction of mRNA secondary structure is critical in designing RNA-based therapeutics as it dictates various steps of an mRNA life cycle, including transcription, translation, and decay. Our application explores how quantum computers could enable scientists to predict mRNA secondary structures at greater scale, accuracy and efficiency than ever before. 

Key Highlights

• Predicting mRNA secondary structures is a formidable combinatorial problem due to the vast number of possible configurations and their associated energy state. Predicting RNA secondary structures with pseudoknots motives is known to be NP-complete. State-of-the-art classical approaches are still constrained by the combinatorial explosion of possible structures, lack of generalization to novel RNAs, and the approximations required to make the problem tractable. As a result, there is a growing interest in leveraging quantum computing to overcome these limitations and achieve more accurate predictions of mRNA secondary structures.
• Value Proposition: We demonstrate that optimization schemes like CVaR-VQE running on noisy quantum hardware with careful calibration and nominal error-mitigation, can produce sufficiently reliable outputs even for relatively large problem instances.
• Advantage vs Classical: Our results indicate that quantum computing offers valuable insights into RNA folding mechanisms and enables rapid identification of functional RNA structures. This work lays the groundwork for future research in quantum-assisted bioinformatics and highlights the importance of developing scalable quantum algorithms for real-world biological problems.
• Availability: As quantum hardware continues to advance, any improvements in algorithmic techniques will further bolster prospects of using quantum computers for optimization. This study, which primarily focused on the capabilities of utility-scale quantum hardware, only considered a subset of available quantum approaches; however, our future work will explore more advanced optimization techniques. 

Detailed Description

Moderna and IBM examine the feasibility of solving the mRNA secondary structure prediction problem on a utility-scale quantum computer with sequence lengths up to 60 nucleotides representing problems in the qubit range of 10 to 80. Our approach utilizes:

• Conditional Value at Risk Variational Quantum Eigensolver (CVaR-VQE), that employs CVaR values of the energy eigenfunctions as the objective function, to achieve better convergence
• Nakanishi-Fujii-Todo (NFT) as a classical optimizer along with a hardware-efficient “two-local” ansatz that is relatively robust against hardware noise and exhibits better convergence properties
• Hardware runs on IBM Quantum Eagle and Heron processors
• To our encouragement, even with “minimal” error mitigation and fixed-depth circuits, our hardware runs yield accurate predictions of minimum free energy (MFE) structures that match the results of the classical solver CPLEX. Our results provide sufficient evidence for the viability of solving mRNA structure prediction problems on a quantum computer and motivate continued research in this direction. 

Publicly Available Technical Data

https://arxiv.org/html/2405.20328v1 

Disclaimer

This Demo is for demonstration purposes only and not for any other purposes. Moderna & IBM own all right, title, and interest in this Demo. Moderna makes no representations or warranties regarding this Demo. To the extent permitted by law, Moderna disclaims all liability including direct and indirect, incidental, or consequential damages, including loss of profits, arising out of or relating to the use of this Demo. The laws designated by the Moderna entity providing this Demo will govern without reference to conflicts of law principles.