A QUBO model of the RNA folding problem optimized by variational hybrid quantum annealing

TL;DR

This paper introduces a QUBO model for RNA folding that leverages quantum annealing and circuit-model quantum computers, aiming to improve prediction accuracy of RNA secondary structures beyond classical heuristics.

Contribution

The authors develop a novel QUBO formulation for RNA folding optimized via variational hybrid quantum annealing, enabling more exact predictions of RNA structures including pseudoknots.

Findings

QUBO model performs comparably to classical methods on known structures

Variational hybrid quantum annealing improves model tuning

Potential for quantum computing to solve complex RNA folding problems

Abstract

RNAs self-interact through hydrogen-bond base-pairing between nucleotides and fold into specific, stable structures that substantially govern their biochemical behaviour. Experimental characterization of these structures remains difficult, hence the desire to predict them computationally from sequence information. However, correctly predicting even the base pairs involved in the folded structure of an RNA, known as secondary structure, from its sequence using minimum free energy models is understood to be NP-hard. Classical approaches rely on heuristics or avoid considering pseudoknots in order to render this problem more tractable, with the cost of inexactness or excluding an entire class of important RNA structures. Given their prospective and demonstrable advantages in certain domains, including combinatorial optimization, quantum computing approaches by contrast have the potential to compute the full RNA folding problem while remaining more feasible and exact. Herein, we present a physically-motivated QUBO model of the RNA folding problem amenable to both quantum annealers and circuit-model quantum computers and compare the performance of this formulation versus current RNA folding QUBOs after tuning the parameters of all against known RNA structures using an approach we call "variational hybrid quantum annealing".