Abstract We develop quantum computational tools to predict the 3D structure of proteins, one of the most important problems in current biochemical research. We explain how to combine recent deep learning advances with the well known technique of quantum walks applied to a Metropolis algorithm. The result, QFold, is a fully scalable hybrid quantum algorithm that, in contrast to previous quantum approaches, does not require a lattice model simplification and instead relies on the much more realistic assumption of parameterization in terms of torsion angles of the amino acids. We compare it with its classical analog for different annealing schedules and find a polynomial quantum advantage, and implement a minimal realization of the quantum Metropolis in IBMQ Casablanca quantum system.
Abstract We develop quantum computational tools to predict the 3D structure of proteins, one of the most important problems in current biochemical research. We explain how to combine recent deep learning advances with the well known technique of quantum walks applied to a Metropolis algorithm. The result, QFold, is a fully scalable hybrid quantum algorithm that, in contrast to previous quantum approaches, does not require a lattice model simplification and instead relies on the much more realistic assumption of parameterization in terms of torsion angles of the amino acids. We compare it with its classical analog for different annealing schedules and find a polynomial quantum advantage, and implement a minimal realization of the quantum Metropolis in IBMQ Casablanca quantum system.
Protein folding is a computationally intensive problem in bioinformatics that involves predicting the three-dimensional structure of a protein from its amino acid sequence. Classical approaches face challenges due to the exponential complexity of the search space. In this paper, we propose Q-ProteinFold, a quantum-classical hybrid system that leverages the Variational Quantum Eigensolver (VQE) for protein folding approximation. The system encodes protein sequences into parameterized quantum circuits using Qiskit and explores multiple conformations through quantum superposition and entanglement. A classical optimizer iteratively minimizes the energy function to identify stable protein structures. The implementation is developed using Python, Streamlit, and Qiskit Aer simulator. Experimental results demonstrate energy convergence, probabilistic state distribution, and effective fold prediction. This work highlights the potential of quantum computing in solving complex biological optimization problems.
Abstract Predicting the three-dimensional structure of a protein from its primary sequence of amino acids is known as the protein folding problem. Due to the central role of proteins’ structures in chemistry, biology and medicine applications, this subject has been intensively studied for over half a century. Although classical algorithms provide practical solutions for the sampling of the conformation space of small proteins, they cannot tackle the intrinsic NP-hard complexity of the problem, even when reduced to the simplest Hydrophobic-Polar model. On the other hand, while fault-tolerant quantum computers are beyond reach for state-of-the-art quantum technologies, there is evidence that quantum algorithms can be successfully used in noisy state-of-the-art quantum computers to accelerate energy optimization in frustrated systems. In this work, we present a model Hamiltonian with O(N4) O ( N 4 ) scaling and a corresponding quantum variational algorithm for the folding of a polymer chain with N monomers on a lattice. The model reflects many physico-chemical properties of the protein, reducing the gap between coarse-grained representations and mere lattice models. In addition, we use a robust and versatile optimization scheme, bringing together variational quantum algorithms specifically adapted to classical cost functions and evolutionary strategies to simulate the folding of the 10 amino acid Angiotensin on 22 qubits. The same method is also successfully applied to the study of the folding of a 7 amino acid neuropeptide using 9 qubits on an IBM 20-qubit quantum computer. Bringing together recent advances in building gate-based quantum computers with noise-tolerant hybrid quantum-classical algorithms, this work paves the way towards accessible and relevant scientific experiments on real quantum processors.
This paper focuses on developing an innovative solution to the protein folding problem, a long-standing challenge in Bioinformatics, through the lens of quantum reinforcement learning (QRL). It proposes a hybrid method that incorporates a Quantum Neural Network created using a Parameterized Quantum Circuit. The study is conducted on the Hydropolar Model, a simplified representation of proteins in terms of their polarity, and the experiments are carried out in both 2D and 3D lattice representations. The datasets used for analysis include benchmark proteins from existing research and PDB or FASTA files from the Protein Data Bank. The core of the paper presents the detailed process of modeling and implementing the protein folding problem using QRL, along with detailed comparisons with classical deep reinforcement models inspired by the literature. The findings from this study demonstrate that QRL performs effectively in predicting energies close to optimal levels. However, it poses a greater challenge in terms of convergence when compared to classical methods. Besides the novel method proposed, we also present a supporting custom tool - the protein-folding-gym-utils python library - designed specifically to assist researchers in applying reinforcement learning techniques to the Protein Folding Problem.
Protein folding is a fundamental biological process in which a linear chain of amino acids spontaneously adopts a unique three-dimensional structure essential for proper biological function, molecular recognition and transport of nutrients. Protein misfolding, on the other hand can lead to severe neurodegenerative disorders such as Alzheimer’s, Parkinson’s and cystic fibrosis. Traditional approaches to protein folding, such as knowledge-based and statistical methods, often face limitations in scalability and physical accuracy. To overcome these challenges, the problem has been reformulated as a combinatorial optimization problem classified as NP-hard. In this work, we implement the CVaR-VQE algorithm to solve a previously proposed QUBO formulation of lattice protein folding within a gate-based quantum computing framework. We demonstrate that, compared to the quantum annealing approach, this method significantly reduces qubit requirements and improves scalability, enabling more efficient and practical implementations on NISQ devices. Additionally, we benchmarked our results against classical optimization using IBM CPLEX, we observed that the proposed quantum approach yields high-quality solutions across multiple peptide instances.
- A study published in the international journal npj Quantum Information (Nature Portfolio) presents a resource-efficient quantum algorithm framework based on the Variational Quantum Eigensolver (VQE) for peptide and protein folding. - By translating protein folding into a tetrahedral lattice model and combining Conditional Value-at-Risk (CVaR) VQE with a classical evolutionary algorithm, the number of required qubits was drastically reduced. - It successfully folded a 7-amino acid neuropeptide using only 9 qubits on actual IBM 20-qubit quantum hardware, demonstrating its practical potential for next-generation drug discovery in the NISQ (Noisy Intermediate-Scale Quantum) era. [Quantum Biology Society] The protein folding problem, which involves predicting the three-dimensional structure of proteins and peptides from their primary amino acid sequence, plays a central role in various applications such as chemistry, biology, and medicine. However, as the sequence lengthens, the number of possible structures to explore explodes exponentially, making it difficult for classical computers to handle its intrinsic NP-hard complexity. Recently, a new quantum algorithm that can efficiently solve this long-standing challenge even on noisy intermediate-scale quantum (NISQ) computers has been developed, drawing the attention of the academic community. The paper titled "Resource-efficient quantum algorithm for protein folding," published in the international journal npj Quantum Information, presents an innovative computing framework for the discovery of next-generation peptide drugs. The research team of Anton Robert and Ivano Tavernelli from IBM Quantum, IBM Research - Zurich in Switzerland successfully led this groundbreaking study. ■ The Meeting of Lattice Models and Qubit Optimization Until now, attempts to introduce quantum computing to protein folding have required an unattainably high number of qubits that current, near-term quantum hardware cannot handle. To solve this, the research team converted protein folding into a tetrahedral lattice model. This model can efficiently depict the growth of a polymer chain while maintaining chemically plausible bond angles. In particular, they adopted a method of adding penalty terms to the Hamiltonian to prevent unphysical amino acid overlaps (e.g., beads occupying the same lattice site). Through this, they succeeded in drastically reducing the number of qubits required for calculation while suppressing the complexity of the algorithm to the level of O(N⁴). ■ Demonstration of Folding on Actual Quantum Hardware This algorithm maximized efficiency by combining CVaR-VQE, a variation of the Variational Quantum Eigensolver (VQE), with a classical Differential Evolution optimizer. Applying this technique, the researchers simulated the folding of the 10-amino acid Angiotensin peptide using only 22 qubits, successfully finding the most stable energy state (native fold). Furthermore, they successfully demonstrated the folding of a 7-amino acid neuropeptide (APRLRFY) using only 9 qubits on IBM's 20-qubit quantum computer hardware. This study is a groundbreaking achievement proving that the protein folding problem, which previously required massive computing resources, can be successfully solved even in a noisy real quantum processor environment. It is expected to bring tremendous innovation to next-generation /de novo/ drug design and disease treatment research utilizing quantum computing in the future.
Protein structure prediction is an important problem in bioinformatics, because the structure of a protein largely determines its biological function. Experimental methods give accurate structures but are expensive and slow, and large deep learning models need heavy computation. This paper presents a hybrid quantum-classical framework that predicts protein secondary structure, a step toward full folding prediction, from amino acid sequences. Sequences from the CB513 benchmark are validated, encoded numerically and reduced to a compact feature vector with principal component analysis (PCA). A Hybrid Quantum Neural Network (HQNN) then processes this vector: a classical encoder transforms the features, a variational quantum circuit with trainable rotation and entangling gates produces new features through measurement, and a classical layer assigns each residue to Helix (H), Sheet (E) or Coil (C). The quantum circuit runs on a simulator, so no quantum hardware is needed. On CB513 the model reaches 87.6% training accuracy and 82.4% validation/test accuracy. A web application takes a sequence as input and returns the predicted structure labels. The results show that a small simulated HQNN can learn the three secondary structure classes; they are not evidence of quantum advantage.