Metalloproteins are a family of proteins characterized by metal ion binding, whereby the presence of these ions confers key catalytic and ligand-binding properties. Due to their ubiquity among biological systems, researchers have made immense efforts to predict the structural and functional roles of metalloproteins. Ultimately, having a comprehensive understanding of metalloproteins will lead to tangible applications, such as designing potent inhibitors in drug discovery. Recently, there has been an acceleration in the number of studies applying machine learning to predict metalloprotein properties, primarily driven by the advent of more sophisticated machine learning algorithms. This review covers how machine learning tools have consolidated and expanded our comprehension of various aspects of metalloproteins (structure, function, stability, ligand-binding interactions, and inhibitors). Future avenues of exploration are also discussed.
Membranes are compelling devices for many industrial separation processes, which are all subject to the intrinsic permeability–selectivity tradeoff. A general strategy to enhance separation performance is to reduce the pore size distribution and, ideally, make the membrane isoporous. In this study, we focus on a minimal model for regularly porous membranes, which consists of hard spheres moving through cylindrical pores. The collision dynamics is solved exactly and implemented in nonequilibrium event-driven molecular dynamics simulations. For such size-sieving porous membranes, we show that the permeability P of hard spheres of size σ through cylindrical pores of size d follows the hindered diffusion mechanism due to size exclusion as P ∝ (1 − σ/d)2. According to this law, the separation of binary mixtures of large and small particles exhibits a linear relationship between α−1/2 and P−1/2, where α and P are the selectivity and permeability of the smaller particle, respectively. The mean permeability through polydisperse pores is the average of the permeabilities of individual pores, weighted by the fraction of the single pore area over the total pore area.
We introduce a variational quantum autoencoder tailored for de novo molecular design named QOBRA (Quantum Operator-Based Real-Amplitude autoencoder). QOBRA le circuits for real-amplitude encoding and the SWAP test to estimate reconstruction and latent-space regularization errors during back-propagation. Adjoint encoderand decoder unitary transformations and a generative process that ensures accurate reconstruction as well as novelty, uniqueness, and validity of the generated samples. We showcase QOBRA as applied to de novo design of Ca 2+ -, Mg 2+ -, and Zn 2+ -binding metalloproteins after training the generative model with a modest dataset. Significance Statement: Recent advancements in classical generative machine learning have shown significant strides in molecular design for targeted applications. N onetheless, these advancements are fundamentally limited by classical computation based on binary units. We introduce a quantum computation-based ML framework employing qubits, which exhibits the ability to synthesize de novo molecular instances with specified properties from limited datasets. Quantum networks require exponentially fewer parameters than classical ones, enhancing their trainability and efficiency. While our demonstration focuses on metalloprotein primary sequences, the paradigm is adaptable to diverse molecular designs. This integration of AI and quantum computing holds potential to expand the scientific and technological frontiers of both domains within a practical framework.
3 February 2026 in Journal of Chemical Theory and Computation