Steric clashes pose a challenge when exploring dense protein systems using conventional explicit-chain methods. A minimal example is a single lattice protein confined on a minimal grid, with no free sites. Finding its minimum energy is a hard optimization problem, with similarities to scheduling problems. It can be recast as a quadratic unconstrained binary optimization (QUBO) problem amenable to classical and quantum approaches. We show that this problem in its QUBO form can be swiftly and consistently solved for chain length 48, using either classical simulated annealing or hybrid quantum-classical annealing on a D-Wave system. In fact, the latter computations required about 10 s. We also test linear and quadratic programming methods, which work well for a lattice gas but struggle with chain constraints. All methods are benchmarked against exact results obtained from exhaustive structure enumeration, at a high computational cost.
Steric clashes pose a challenge when exploring dense protein systems using conventional explicit-chain methods. A minimal example is a single lattice protein confined on a minimal grid, with no free sites. Finding its minimum energy is a hard optimization problem, with similarities to scheduling problems. It can be recast as a quadratic unconstrained binary optimization (QUBO) problem amenable to classical and quantum approaches. We show that this problem in its QUBO form can be swiftly and consistently solved for chain length 48, using either classical simulated annealing or hybrid quantum-classical annealing on a D-Wave system. In fact, the latter computations required about 10 s. We also test linear and quadratic programming methods, which work well for a lattice gas but struggle with chain constraints. All methods are benchmarked against exact results obtained from exhaustive structure enumeration, at a high computational cost.
Quantum annealing is a promising approach for obtaining good approximate solutions to difficult optimization problems.Folding a protein sequence into its minimum-energy structure represents such a problem.For testing new algorithms and technologies for this task, the minimal lattice-based [hydrophobic (H) or polar (P) beads] HP model is well suited, as it represents a considerable challenge despite its simplicity.The HP model has favorable interactions between adjacent, not directly bound hydrophobic residues.Here, we develop a novel spin representation for lattice protein folding tailored for quantum annealing.With a distributed encoding onto the lattice, it differs from earlier attempts to fold lattice proteins on quantum annealers, which were based upon chain growth techniques.With our encoding, the Hamiltonian by design has the quadratic structure required for calculations on an Ising-type annealer, without having to introduce any auxiliary spin variables.This property greatly facilitates the study of long chains.The approach is robust to changes in the parameters required to constrain the spin system to chainlike configurations, and performs very well in terms of solution quality.The results are evaluated against existing exact results for HP chains with up to N = 30 beads with 100% hit rate, thereby also outperforming classical simulated annealing.In addition, the method allows us to recover the lowest known energies for N = 48 and N = 64 HP chains, with similar hit rates.These results are obtained by the commonly used hybrid quantum-classical approach.For pure quantum annealing, our method successfully folds an N = 14 HP chain.The calculations were performed on a D-Wave Advantage quantum annealer.
Quantum computing has progressed from a theoretical concept to a practical experimental technology. By processing information according to quantum mechanics rather than classical rules, quantum computers may offer advantages for certain problems. Although a few algorithms have proven speedups, it remains unclear where quantum computers will be practically useful, especially since current devices are noisy, small, and not yet fault tolerant. This thesis explores quantum computing for biophysically motivated combinatorial problems, with a focus on lattice protein models. In these simplified models, a protein is represented as a chain on a discrete lattice. Despite their simplicity, lattice proteins retain important features of real proteins, such as such as the folding of a chain with exponentially many possible shapes to a unique minimum-energy structure and steric constraints.To run lattice protein problems on quantum hardware, they must first be formulated using binary variables and objective functions. The latter typically contain soft constraints needed to ensure physical solutions. In Papers~\I--\II, we developed and extended a coordinate-based mapping for lattice protein folding, enabling D-Wave's hybrid quantum-classical annealer to fold both 2D hydrophobic-polar and 3D Miyazawa-Jernigan lattice proteins. In Paper~\III, we adapted the approach to the gate-based paradigm and showed how quadratic soft constraints could be removed by a mixer developed for the maximum-independent-set problem. Papers~\IV--\V addressed lattice protein design using quantum annealing, quantum approximate optimization algorithm, quantum alternating operator ansatz and hardware-efficient ansatz, while Paper~\VI extended the framework to multi-chain thermodynamics and the reconstruction of densities of states and heat-capacity features.Together, these papers show how folding, design, and multi-chain thermodynamics can be formulated, implemented, and analyzed using present-day quantum and hybrid hardware. A recurring limitation is the treatment of soft constraints. In quantum annealing, their high connectivity leads to error accumulation. On gate-based hardware, they increase circuit depth and sensitivity to noise. At the same time, temporary constraint violations can provide useful freedom during optimization. Further progress will therefore require improved formulations, algorithms and hardware that reduce the cost of enforcing physical solutions while retaining this flexibility.
Identifying minimum-energy structures of lattice proteins is a challenging discrete optimization problem. Quantum approaches such as analog quantum annealing and the gate-based quantum approximate optimization algorithm (QAOA) can address this problem after mapping it to a binary representation, which typically involves introducing penalty terms to enforce valid chain configurations. However, in this and many related problems, the use of quadratic penalty terms can be avoided by restricting the search space to independent sets in a conflict graph and using a QAOA mixer designed for the maximum independent set problem. In this work, we implement and explore this QAOA variant for lattice protein folding. Here, the objective function consists solely of the protein energy together with a simple linear bias term, without quadratic penalties. We validate this approach through classical simulations of the quantum circuits for lattice proteins of lengths N=4 and N=6. To explore larger systems, we further introduce a heuristic iterative local-search scheme, with which we successfully fold lattice proteins with lengths up to N=14 using local subgraphs with at most 26 qubits.
This chapter presents a general construction of the free energy function for the two-dimensional lattice heteropolymer model widely used to study the dynamics of proteins. It demonstrates how to map the lattice heteropolymer problem into forms that can be solved by using different types of technology and algorithms. The embedding strategies presented here apply to many discrete optimization problems. Mapping these problems to a constraint programming problem is a three-step process. The chapter provides a brief description of the process and expands upon each step as it applies to lattice folding. It explains the “circuit” construction that provides optimal efficiency at the cost of introducing high-ordered many-body terms. Finally, the chapter demonstrates the final steps required to embed a small instance of a particular lattice protein problem into a quadratic unconstrained binary optimization (QUBO) Hamiltonian.
Abstract Predicting a protein’s three-dimensional structure from its primary amino acid sequence constitutes the protein folding problem, a pivotal challenge within computational biology. This task has been identified as a fitting domain for applying quantum annealing, an algorithmic technique posited to be faster than its classical counterparts. Nevertheless, the utility of quantum annealing is intrinsically contingent upon the spectral gap associated with the Hamiltonian of lattice proteins. This critical dependence introduces a limitation to the efficacy of these techniques, particularly in the context of simulating the intricate folding processes of proteins. In this paper, we address lattice protein folding as a polynomial unconstrained binary optimization problem, devising a hybrid quantum–classical algorithm to determine the minimum energy conformation effectively. Our method is distinguished by its logarithmic scaling with the spectral gap, conferring a significant edge over the conventional quantum annealing algorithms. The present findings indicate that the folding of lattice proteins can be achieved with a resource consumption that is polynomial in the lattice protein length, provided an ansatz state that encodes the target conformation is utilized. We also provide a simple and scalable method for preparing such states and further explore the adaptation of our method for extension to off-lattice protein models. This work paves a new avenue for surmounting complex computational biology problems via the utilization of quantum computers.
Hybrid quantum-classical annealing solvers are moving from research laboratories toward industrial use, but their advantage over classical optimization appears to be confined to a narrow class of problems. A recent systematic benchmark of the D-Wave hybrid solver against industry-leading classical solvers, across several classes of optimization problems, found it to be most competitive on binary quadratic programs, while noting that few industrial problems take this form. Therefore, identifying real-world industrial problems that can be modeled with this structure is a key challenge. We propose protein side-chain optimization, a core step of computational protein design, as one of these problems: it is natively a binary quadratic program, and its application context requires pools of low-energy solutions, which annealers naturally return. In this feasibility study, we test whether the competitiveness of the hybrid solver against industry-leading classical solvers extends to this problem. On 40 instances derived from 20 experimentally determined protein structures with up to about 200 amino acids and 2000 binary variables, we compare the results achieved with a hybrid solver, run for 5 seconds, against those of the commercial solver Gurobi, run for 10 seconds, 1000 seconds, and 4 hours. The hybrid solver reaches the certified optimum whenever one is available, matches Gurobi's performance on smaller instances, and outperforms it on larger instances, finding lower-energy solutions. These results show that the competitiveness of the hybrid solver on binary quadratic programs, so far observed only on a synthetic benchmark, extends to a real-world problem. This motivates comparisons with classical heuristics specific to protein side-chain optimization.