Determining mutant protein structures is critical for understanding the mechanistic roles of mutations in biochemical processes. However, experimental characterization and conventional theoretical modeling are often expensive and time-consuming. Recent advances in machine learning provide new opportunities to efficiently predict protein structures from primary sequences. Nevertheless, applying these models to proteins with single-site or few-site mutations remains challenging because mutant sequences are often highly similar to their wild-type counterparts. Here, we introduce DeltaDiff, a physics-guided inference framework for mutant-structure generation that incorporates mutation-aware physical guidance into a baseline diffusion model. We evaluate DeltaDiff on three representative systems: Chignolin T8P, Novispirin G-10, and BBL D162N. All three examples involve nonlocal structural changes, making accurate mutant-structure prediction challenging. DeltaDiff captures key mutation-induced conformational changes without requiring retraining or fine-tuning of the baseline model. These results establish a foundation for efficient mutant-structure prediction at a fraction of the cost of conventional methods, facilitating rational mutant design.

In this study, we demonstrate that it is possible to fit the resilience at the high temperature limit using only the atomic mean square displacement determined from cryogenic temperature single crystal diffraction data. Our method introduces the Debye phonon model, under which the atomic mean square displacement displays quantum behavior at cryogenic temperatures and deviates from a linear temperature dependence. From the resilience of Si extrapolated from cryogenic temperature X-ray diffraction data, one can calculate the Si isotope fractionation using our recently developed force constants approach. We applied our method to epidote and coesite, which are common minerals in ultra-high pressure metamorphic rocks, and our predicted Si isotope fractionation ln alpha Si30/28 are consistent with mass spectroscopy observations of Delta 30Si in Dabie eclogite and Alps whiteschist. Our results indicate that the geofluid involved in the fluid-rock interaction in Dabie and Alps orogens doesn't significantly alter the Si-isotope composition of ultrahigh pressure metamorphism. Our manuscript presents a novel experimental methodology to determine the high-temperature resilience and equilibrium isotope fractionation factors of silicon in minerals, which can be applied to the metastable upper mantle silicates that are non-quenchable at high temperatures and room pressure.

Molecular dynamics (MD) simulation has long been the principal computational tool for exploring protein conformational landscapes, but its application is limited by high computational cost. We present ProTDyn, a foundation protein language model that unifies conformational ensemble generation and multi-timescale dynamics modeling within a single framework. Unlike prior approaches that treat these tasks separately, ProTDyn allows flexible i.i.d ensemble sampling and dynamic trajectory simulation. Across diverse protein systems, ProTDyn yields thermodynamically consistent ensembles, faithfully reproduces dynamical properties over multiple timescales, and generalizes to proteins beyond its training data—offering a scalable and efficient alternative to conventional MD simulations.

Complex chemical systems often contain multiple components or large molecules, giving rise to substantial chemical and conformational complexity. Modeling the structures of these systems is crucial for mechanistic understanding of chemical processes and rational design, but remains highly challenging for both conventional theoretical approaches and machine learning. Here, we introduce physical-interaction-coupled diffusion models (PICDiff), a framework that couples independently trained diffusion models for subsystems of a complex system through explicit bonded and nonbonded interactions during inference. PICDiff reduces the difficulty of applying generative models to chemically and conformationally complex systems by decomposing them into smaller subsystems that are less complex and more amenable to machine-learning-based modeling. Using peptide-polymer binding and polymer conformation sampling as examples, we demonstrate that PICDiff can quantitatively sample conformations and model the thermodynamics of complex chemical systems. These results show that PICDiff provides a general and practical approach for modeling complex chemical systems by combining learned models of simpler molecular subsystems through physical interactions.

A bstract Molecular dynamics (MD) simulations remain the standard tool for characterizing protein conformational landscapes, but their high computational cost limits large-scale and long-timescale applications. Recent generative models, especially diffusion-based approaches, provide promising alternatives by learning equilibrium conformational distributions across diverse protein systems. We present ConformFlow , the first scalable normalizing-flow framework for sequence-conditioned protein conformational ensemble generation. ConformFlow combines a continuous backbone latent representation with a RealNVP-style flow parameterized by sequence-aware Transformer coupling networks, enabling exact likelihood training, single-step sampling, and plug-and-play conditioning on flexible geometric constraints. Across diverse protein systems, ConformFlow generates ensembles that agree well with reference MD simulations, generalizes to proteins beyond its training data, and achieves substantially faster sampling than diffusion-based baselines. These results establish ConformFlow as an efficient and controllable alternative for protein conformational ensemble generation. Code https://github.com/Harrydirk41/ConformFlow.git

Using our recently developed X‐ray diffraction based force constants approach, we have determined the equilibrium Si isotope fractionation between omphacite/garnet, quartz/kyanite, and quartz/zircon at temperatures relevant to the petrogenesis. We find that Na strongly affects the Si isotope fractionation between omphacite and garnet. Our results have suggested that the omphacite and garnet in eclogite collected in the Dabie Mountain, as well as the kyanite and its host quartz veins, are isotopically in equilibrium, which further suggests that the Dabie Mountain eclogites and its host veins underwent the same high pressure‐temperature condition during their formation. The Si isotope fractionation determined by our methods, together with published mass spectroscopy measurements, DFT‐CIPW calculations and sigmoid fitting on various felsic granites, have suggested that the Si isotope fraction between zircon and whole rock “saturates” at ∼0.45‰ at 1000 K when the SiO 2 content in the granite is above ∼70 wt%.

In recent years, many foundation generative models have been developed to predict structures of molecules and materials. Although these foundation models have achieved great success, it is challenging to collect enough data to train foundation generative models. One such example is to predict protein conformations with protein-environment interactions (PEIs), such as interactions introduced by organic linkers or material surfaces. We propose a physics-guided route to extrapolate foundation models beyond their training domain. Our method couples a pretrained deep generative model with explicit, physics-based interaction potentials for PEIs, steering sampling toward conformations consistent with external constraints without any retraining or fine-tuning. We demonstrate accurate and efficient conformation prediction of (i) cyclic peptide with organic linkers and (ii) peptide adsorbed on the gold surface. The generated structures serve as high-quality initial conditions for downstream simulations, providing a general, systematic approach to extend foundation models to proteins under system-specific environmental interactions.

Proteins often adopt multiple ensemble conformations to perform essential functions such as catalysis, transport, and signal transduction. Traditional physics-based methods for generating these conformations, including molecular dynamics and Monte Carlo simulations, are computationally expensive and time-consuming, limiting their practicality for high-throughput applications like screening. Recent advances in machine learning, particularly deep generative models, offer a promising alternative for protein conformation ensemble generation. However, these models are often task-specific or rely on strong assumptions to generalize. Here, we introduce ESMAdam, a versatile and efficient framework for protein conformation ensemble generation. Using the ESMFold protein language model ESMFold and ADAM stochastic optimization in the continuous protein embedding space, ESMAdam addresses a wide range of ensemble generation tasks. In this work, we demonstrate several basic applications of ESMAdam, including conditional ensemble generation and CG-to-all-atom backmapping. In addition, we showcase advanced applications, such as screening alternative binding modes of protein multimers and reconstructing 3D structures from cryo-EM images. Compared to traditional physics-based methods, ESMAdam significantly reduces computational time. Unlike deep-generative-model-based approaches, it requires no retraining and easily adapts to diverse ensemble restraint conditions, making it exceptionally suited for various structure prediction and screening tasks. This plug-and-play framework represents a step toward efficient and flexible protein ensemble generation for applications in structural biology and drug discovery. ### Competing Interest Statement The authors have declared no competing interest.

Proteins play crucial roles in life processes such as catalysis, signal transduction, immune response, and molecular transport. Many of these functions are mediated by external interactions between a protein and other chemical species, including proteins, nucleic acids, and organic molecules. Furthermore, in therapeutic applications, protein functions can be influenced by interactions with unnatural chemical species, such as molecular organic linkers, inorganic materials, and polymers. Predicting protein conformational ensembles under the influence of external interactions remains a significant challenge both experimentally and computationally. While deep generative models have been developed to predict protein structural ensembles, their application to systems involving external interactions remains largely unexplored. In this work, we present a method that integrates a deep generative model with physics-based interaction modeling to predict protein conformations under external constraints. Without requiring any retraining or fine-tuning of the generative model, our approach can efficiently and accurately predict protein conformations covalently constrained by organic linkers, as well as protein conformations adsorbed on gold surfaces. ### Competing Interest Statement The authors have declared no competing interest. American Chemical Society Petroleum Research FundAmerican Chemical Society Petroleum Research Fund, , 67307

Understanding protein conformation is key to understanding their function. Importantly, most proteins adopt multiple conformations with non-trivial ensemble distributions that change depending on their environment to perform functions like catalysis, signaling, and transport. Recently, machine learning techniques, especially deep generative models, have been employed to develop protein conformation generators. These models, known as unified protein ensemble samplers, are trained on the PDB dataset and can generate diverse protein conformation ensembles given a protein sequence. However, their reliance solely on structural data from the PDB, which primarily captures folded protein states, restricts the diversity of the generated ensembles and can result in physically unrealistic conformations. In this paper, we overcome these challenges by introducing EGDiff, an experiment-guided diffusion model for protein conformation generation. EGDiff integrates experimental measurements as a physical prior, enabling the generation of protein conformations with desired properties. Our experiments on a variety of fast-folding and intrinsically disordered proteins demonstrate that EGDiff significantly advances the capabilities of current unified protein ensemble samplers. With little computational cost, EGDiff can capture important proteins' configuration properties and the underlying Boltzmann distribution, paving the way for a next-generation molecular dynamics engine. We further demonstrate the effectiveness of EGDiff to capture conformational changes in the presence of mutations and as an efficient tool for determining a reasonable CV space for protein ensembles. With these results, EGDiff is well-poised to push the study of protein ensembles into a data-rich regime currently available to few problems in biology. ### Competing Interest Statement The authors have declared no competing interest.

Molecular polaritons arise when molecules interact so strongly with light that they become entangled with each other. This light–matter hybridization alters the chemical and physical properties of the molecular system and allows chemical reactions to be controlled without the use of external fields. We investigate the impact of strong light–matter coupling on the electronic structure using perturbative approaches and demonstrate that Rayleigh–Schrödinger perturbation theory can reproduce the ground state energies in optical cavities to comparable accuracy as ab initio cavity quantum electrodynamics methodologies for currently relevant coupling strengths. The method is effective in both low and high cavity frequency regimes and straightforward to implement via response functions. Furthermore, we establish simple relations between cavity-induced intermolecular forces and van der Waals forces. These findings provide valuable insight into the manipulation of ground-state polaritonic energy landscapes, shedding light on the systems and conditions in which modifications can be achieved.

Metal organic chalcogenides (MOCs) are an emerging class of two-dimensional (2D) materials featuring tunable band gaps and strong light-matter interactions with great potential for optical and polaritonic applications. Lead organic chalcogenides (LOCs) stand out among MOCs for sustaining long-lived coherent optical phonons despite their distorted lattices. However, the strong electron-phonon coupling also leads to rapid charge carrier self-trapping, hindering carrier transport. Herein, we demonstrate that charge carrier self-trapping is suppressed in Se- and Te-based LOCs. Specifically, the expansive Se and Te orbitals give rise to more dispersive conduction bands and reduced electron effective masses, thereby mitigating carrier self-trapping. As a result of reduced carrier self-trapping, the Se- and Te-based LOCs exhibit enhanced band-to-band photoluminescence and improved charge transport performance. Our results provide a synthetic route to Se- and Te-based LOC single crystals and demonstrate the potential of orbital engineering to tune their electronic and phononic properties.

Biased enhanced sampling methods that utilize collective variables (CVs) are powerful tools for sampling conformational ensembles. Due to their large intrinsic dimensions, efficiently generating conformational ensembles for complex systems requires enhanced sampling on high-dimensional free energy surfaces. While temperature-accelerated molecular dynamics (TAMD) can trivially adopt many CVs in a simulation, unbiasing the simulation to generate unbiased conformational ensembles requires accurate modeling of a high-dimensional CV probability distribution, which is challenging for traditional density estimation techniques. Here we propose an unbiasing method based on the score-based diffusion model, a deep generative learning method that excels in density estimation across complex data landscapes. We demonstrate that this unbiasing approach, tested on multiple TAMD simulations, significantly outperforms traditional unbiasing methods and can generate accurate unbiased conformational ensembles. With the proposed approach, TAMD can adopt CVs that focus on improving sampling efficiency and the proposed unbiasing method enables accurate evaluation of ensemble averages of important chemical features.

Aqueous zinc metal batteries (AZMB) are emerging as a promising alternative to the prevailing existing Lithium-ion battery technology. However, the development of AZMBs is hindered due to challenges including dendrite formation, hydrogen evolution reaction (HER), and ZnO passivation on the anode. Here, a tetraalkylsulfonamide (TAS) additive for suppressing HER, dendrite formation, and enhancing cyclability is rationally designed. Only 1 mm TAS is found that can effectively displace water molecules from the Zn2+ solvation shell, thereby altering the solvation matrix of Zn2+ and disrupting the hydrogen bond network of free water, as demonstrated through (67) Zn and H-1 nuclear magnetic resonance spectroscopy, high-resolution mass spectrometry (HRMS), and density functional theory (DFT) studies. Voltammetry synchronized with in situ monitoring of the electrode surface reveals suppressed dendritic growth and HER in the presence of TAS. Electrochemical mass spectrometry (ECMS) captures real-time HER suppression during Zn electrodeposition, revealing the ability of TAS to suppress the HER by an order of magnitude. A approximate to 25-fold cycle life improvement from approximate to 100 h to over 2500 h in coin cells cycled in the presence of TAS. Furthermore, by suppressing passivation product formation, it is demonstrated that strategy robustly maximizes the stability of Zn metal anodes.

Hybrid organic-inorganic semiconductors with strong electron-phonon interactions provide a programmable platform for developing a variety of electronic, optoelectronic, and quantum materials by controlling these interactions. However, in current hybrid semiconductors such as halide perovskites, anharmonic vibrations with rapid dephasing hinder the ability to coherently manipulate phonons. Here, we report the observation of long-lived coherent phonons in lead organic chalcogenides (LOCs), a new family of hybrid two-dimensional semiconductors. These materials feature harmonic phonon dynamics despite distorted lattices, combining long phonon dephasing times with tunable semiconducting properties. A dephasing time -up to 75 ps at 10 K, with up to ∼500 cycles of phonon oscillation between scattering events, was observed, corresponding to a dimensionless harmonicity parameter that is more than an order of magnitude larger than that of halide perovskites. The phonon dephasing time is significantly influenced by anharmonicity and centrosymmetry, both of which can be tuned through the design of the organic ligands enabled by the direct bonding between the organic and inorganic motifs. This research opens new opportunities for the manipulation of electronic properties with coherent phonons in hybrid semiconductors.

Gradient-structured alloys with lamellar microstructures show extraordinary potential in breaking the strength-ductility trade-off dilemma. When partial grain size exceeds the Hall-Petch limit, the deformation mechanism is not fully understood. In this study, the competing relationship between grain boundary (GB) softening and hetero-deformation-induced (HDI) strengthening in laminated nano-grained aluminum was investigated under uniaxial tensile loading. In the soft domains, the high activity of grain boundaries results in pronounced GB softening. However, when the volume fraction of the soft layer exceeds 50%, the flow stress surpasses the predicted values of the rule of mixtures (ROM), demonstrating a significant HDI strengthening. A detailed analysis of microstructural evolution is conducted, including strain gradients, GB migration, and dislocation distribution. The effects of mechanical incompatibility between soft and hard layers on the distribution of plastic strain and overall material strength are elucidated in gradient structures. The study provides critical theoretical insights into the design and development of high-performance gradient nanocrystalline aluminum alloys, particularly in applications where a balance between strength and ductility is of paramount importance.

Cesium lead bromide (CsPbBr3) is a prominent halide perovskite with extensive optoelectronic applications. In this study, we report the pressure modulation of CsPbBr3's crystal structure and electronic properties at room temperature up to 5 GPa. We have observed a crystal structure transition from the orthorhombic Pnma space group to a new monoclinic phase in the space group P2(1)/c at 2.08 GPa. The structure is associated with similar to 8% of density jump across the transition boundary. DFT calculations have suggested that the structure transition leads to a change in the electronic band structure, and there is an emergent indirect bandgap at the Pnma-P2(1)/c phase transition boundary at 2.08 GPa. Across the transition boundary, the electronic band gap of CsPbBr3 increased from 2.07 eV to 2.38 eV, which explains its pressure-induced color change. Our study demonstrates the importance of using in-situ crystal structure in the electronic band structure calculations in halide perovskites.

Density Functional Theory (DFT) has become a cornerstone in the modeling of metals. However, accurately simulating metals, particularly under extreme conditions, presents two significant challenges. First, simulating complex metallic systems at low electron temperatures is difficult due to their highly delocalized density matrix. Second, modeling metallic warm-dense materials at very high electron temperatures is challenging because it requires the computation of a large number of partially occupied orbitals. This study demonstrates that both challenges can be effectively addressed using the latest advances in linear-scaling stochastic DFT methodologies. Despite the inherent introduction of noise into all computed properties by stochastic DFT, this research evaluates the efficacy of various noise reduction techniques under different thermal conditions. Our observations indicate that the effectiveness of noise reduction strategies varies significantly with the electron temperature. Furthermore, we provide evidence that the computational cost of stochastic DFT methods scales linearly with system size for metal systems, regardless of the electron temperature regime.

Coarse-grained (CG) models play a crucial role in the study of protein structures, protein thermodynamic properties, and protein conformation dynamics. Due to the information loss in the coarse-graining process, backmapping from CG to all-atom configurations is essential in many protein design and drug discovery applications when detailed atomic representations are needed for in-depth studies. Despite recent progress in data-driven backmapping approaches, devising a backmapping method that can be universally applied across various CG models and proteins remains unresolved. In this work, we propose BackDiff, a new generative model designed to achieve generalization and reliability in the protein backmapping problem. BackDiff leverages the conditional score-based diffusion model with geometric representations. Since different CG models can contain different coarse-grained sites which include selected atoms (CG atoms) and simple CG auxiliary functions of atomistic coordinates (CG auxiliary variables), we design a self-supervised training framework to adapt to different CG atoms, and constrain the diffusion sampling paths with arbitrary CG auxiliary variables as conditions. Our method facilitates end-to-end training and allows efficient sampling across different proteins and diverse CG models without the need for retraining. Comprehensive experiments over multiple popular CG models demonstrate BackDiff's superior performance to existing state-of-the-art approaches, and generalization and flexibility that these approaches cannot achieve. A pretrained BackDiff model can offer a convenient yet reliable plug-and-play solution for protein researchers, enabling them to investigate further from their own CG models.

Intermolecular van der Waals interactions are central to chemical and physical phenomena ranging from biomolecule binding to soft-matter phase transitions. In this work, we demonstrate that strong light-matter coupling can be used to control the thermodynamic properties of many-molecule systems. Our analyses reveal orientation dependent single molecule energies and interaction energies for van der Waals molecules. For example, we find intermolecular interactions that depend on the distance between the molecules R as R-3 and R0. Moreover, we employ ab initio cavity quantum electrodynamics calculations to develop machine-learning-based interaction potentials for molecules inside optical cavities. By simulating systems ranging from 12 H2 to 144 H2 molecules, we observe varying degrees of orientational order because of cavity-modified interactions, and we explain how quantum nuclear effects, light-matter coupling strengths, number of cavity modes, molecular anisotropies, and system size all impact the extent of orientational order.