Protein engineering in the deep learning era.

Advances in deep learning have significantly aided protein engineering in addressing challenges in industrial production, healthcare, and environmental sustainability. This review frames frequently researched problems in protein understanding and engineering from the perspective of deep learning. It provides a thorough discussion of representation methods for protein sequences and structures, along with general encoding pipelines that support both pre-training and supervised learning tasks.

Switchable Acidic Oxygen Evolution Mechanisms on Atomic Skin of Ruthenium Metallene Oxides.

RuO has been considered as a promising, low-cost, and highly efficient catalyst in the acidic oxygen evolution reaction (OER). However, it suffers from poor stability due to the inevitable involvement of the lattice oxygen mechanism (LOM). Here, we construct a unique metallene-based core-skin structure and unveil that the OER pathway of atomic RuO skin can be regulated from the LOM to an adsorbate evolution mechanism by altering the core species from metallene oxides to metallenes.

Hydroxyl Spillover in Fe-Se Dual-Site Catalysts for Mixed Plastics Assay.

The complex composition of real plastic wastes poses a significant challenge for their large-scale disposal. A responsive on-site compositional analysis of plastics is informative in choosing downstream processing methods. Nanocatalyst-based assay kit is highly qualified for this scene; however, there remain no efficient nanocatalysts for plastics due to their highly inert chemistry.

Ru-OH-Zr Site over Metal-Organic Frameworks Boosts Coreactant Activation for Efficient Electrochemiluminescence.

Metal-organic frameworks (MOFs) are promising electrochemiluminescent (ECL) nanoemitters. Great endeavors have been made to optimize the inherent luminescent properties, yet most MOFs suffer from poor coreactant activation ability, resulting in limited ECL. Therefore, it is urgent to integrate and design efficient catalytic centers within MOFs.

Data-driven natural computational psychophysiology in class.

The assessment of mental fatigue (MF) and attention span in educational and healthcare settings frequently relies on subjective scales or methods such as induced-task interruption tools. However, these approaches are deficient in real-time evaluation and dynamic definitions. To address this gap, this paper proposes a Continuous Quantitative Scale (CQS) that allows for the natural and real-time measurement of MF based on group-synchronized electroencephalogram (EEG) data.

Depth-of-Discharge Dependent Capacity Decay Induced by the Accumulation of Oxidized Lattice Oxygen in Li-Rich Layered Oxide Cathode.

More and more basic practical application scenarios have been gradually ignored/disregarded, in fundamental research on rechargeable batteries, e.g. assessing cycle life under various depths-of-discharge (DODs).

A general temperature-guided language model to design proteins of enhanced stability and activity.

Designing protein mutants with both high stability and activity is a critical yet challenging task in protein engineering. Here, we introduce PRIME, a deep learning model, which can suggest protein mutants with improved stability and activity without any prior experimental mutagenesis data for the specified protein. Leveraging temperature-aware language modeling, PRIME demonstrated superior predictive ability compared to current state-of-the-art models on the public mutagenesis dataset across 283 protein assays.

Major-auxiliary cooperative metal pairs in MOFs enable cascade oxidation of KA oil to ε-caprolactone.

Direct oxidation of KA oil (the mixture of cyclohexanone and cyclohexanol) toward ε-caprolactone is in high demand yet hard to implement in need of juggling the activation of both methyne C-H bond of cyclohexanol and α-C-C bond of cyclohexanone. Here we demonstrate that in situ formed Cu-O active site, which originates from relay reaction at Ni(II) and Cu(I) pairs in a metal-organic framework (known as NiCu-MOF-74) with O and benzaldehyde (PhCHO), efficiently oxidizes KA oil toward ɛ-caprolactone along with good stability. Mechanism investigation discloses that the auxiliary Ni(II) site first adsorbs O for abstracting formyl hydrogen in PhCHO followed by transfer of PhCO· to react with another O over the major Cu(I) site, leading to formation of Cu-O and PhCOOH.

Local compressive strain-induced anti-corrosion over isolated Ru-decorated CoO for efficient acidic oxygen evolution.

Enhancing corrosion resistance is essential for developing efficient electrocatalysts for acidic oxygen evolution reaction (OER). Herein, we report the strategic manipulation of the local compressive strain to reinforce the anti-corrosion properties of the non-precious CoO support. The incorporation of Ru single atoms, larger in atomic size than Co, into the CoO lattice (Ru-CoO), triggers localized strain compression and lattice distortion on the Co-O lattice.

Electronic Modulation and Symmetry-Breaking Engineering of Single-Atom Catalysts Driving Long-Cycling Li-S Battery.

Developing efficient and durable single-atom catalysts is vitally important for the sulfur redox reaction (SROR) in Li-S battery, while it remains enormous challenging. Herein, undercoordinated Ni-N moieties anchored on N,S-codoped porous carbon (Ni-NSC) is obtained to enhance the SROR. The experiments and theoretical calculations indicate that the symmetry-breaking charge transfer in Ni single-atom catalyst originates from tuning effect of sulfur atoms mediated Ni-N moieties, which can both facilitate the chemical adsorption by formation of N-Ni⋅⋅⋅S , and achieve a rapid redox conversion of polysulfides because of the enhanced electron transfer.

Single-Atom Pt Loaded on MOF-Derived Porous TiO with Maxim-Ized Pt Atom Utilization for Selective Hydrogenation of Halonitro-benzene.

The location control of single atoms relative to supports is challenging for single-atom catalysts, leading to a large proportion of inaccessible single atoms buried under supports. Herein, a "sequential thermal transition" strategy is developed to afford single-atom Pt preferentially dispersed on the outer surface of TiO. Specifically, a Ti-MOF confining Pt nanoparticles is converted to Pt and TiO composite coated by carbon (Pt&TiO@C-800) at 800 °C in N.

Evolution of Nitrogen-Coordinated Metal Single Atoms Toward Single-Atom Alloys on MgH₂ as Efficient and Stable Hydrogen Spillover Catalysts.

M-N-C catalysts with nitrogen-coordinated metal single-atom active sites have demonstrated high activity for hydrogen storage materials, but their stability in this application remains uncertain. This study addresses this issue by using nickel phthalocyanine (NiPc) molecules on MgH₂ particles as a model system. It is found that the N-coordinated high-valence Ni single atoms in the NiN₄ active site are unstable in the reducing environment of hydrogen storage, spontaneously evolving into zero-valence Ni, forming a Ni₁-Mg single-atom alloy (SAA).

Switching CO Electroreduction Pathways between Ethylene and Ethanol via Tuning Microenvironment of the Coating on Copper Nanofibers.

Engineering the microenvironment of electrode surface is one of the effective means to tune the reaction pathways in CORR. In this work, we prepared copper nanofibers with conductive polypyrrole coating by polymerization of pyrrole using polyvinyl pyrrolidone (PVP) as template. As a result, the obtained copper nanofibers Cu/CuO/SHNC, exhibited a superhydrophobic surface, which demonstrated very high selectivity for ethanol with a Faraday efficiency (FE) of 66.

A conditional protein diffusion model generates artificial programmable endonuclease sequences with enhanced activity.

Deep learning-based methods for generating functional proteins address the growing need for novel biocatalysts, allowing for precise tailoring of functionalities to meet specific requirements. This advancement leads to the development of highly efficient and specialized proteins with diverse applications across scientific, technological, and biomedical fields. This study establishes a pipeline for protein sequence generation with a conditional protein diffusion model, namely CPDiffusion, to create diverse sequences of proteins with enhanced functions.

Fullerene on non-iron cluster-matrix co-catalysts promotes collaborative H and N activation for ammonia synthesis.

Developing highly effective catalysts for ammonia (NH) synthesis is a challenging task. Even the current, prevalent iron-derived catalysts used for industrial NH synthesis require harsh reaction conditions and involve massive energy consumption. Here we show that anchoring buckminsterfullerene (C) onto non-iron transition metals yields cluster-matrix co-catalysts that are highly efficient for NH synthesis.

Scalable Multi-FPGA HPC Architecture for Associative Memory System.

Associative memory is a cornerstone of cognitive intelligence within the human brain. The Bayesian confidence propagation neural network (BCPNN), a cortex-inspired model with high biological plausibility, has proven effective in emulating high-level cognitive functions like associative memory. However, the current approach using GPUs to simulate BCPNN-based associative memory tasks encounters challenges in latency and power efficiency as the model size scales.

Decoupling of the onset of anharmonicity between a protein and its surface water around 200 K.

The protein dynamical transition at ~200 K, where the biomolecule transforms from a harmonic, non-functional form to an anharmonic, functional state, has been thought to be slaved to the thermal activation of dynamics in its surface hydration water. Here, by selectively probing the dynamics of protein and hydration water using elastic neutron scattering and isotopic labeling, we found that the onset of anharmonicity in the two components around 200 K is decoupled. The one in protein is an intrinsic transition, whose characteristic temperature is independent of the instrumental resolution time, but varies with the biomolecular structure and the amount of hydration, while the one of water is merely a resolution effect.

Unveiling the Activity Origin of Electrochemical Oxygen Evolution on Heteroatom-Decorated Carbon Matrix.

Chemical modification via functional dopants in carbon materials holds great promise for elevating catalytic activity and stability. To gain comprehensive insights into the pivotal mechanisms and establish structure-performance relationships, especially concerning the roles of dopants, remains a pressing need. Herein, we employ computational simulations to unravel the catalytic function of heteroatoms in the acidic oxygen evolution reaction (OER), focusing on a physical model of high-electronegative F and N co-doped carbon matrix.

Precious-Metal-Free Mo-MXene Catalyst Enabling Facile Ammonia Synthesis Via Dual Sites Bridged by H-Spillover.

To date, NH synthesis under mild conditions is largely confined to precious Ru catalysts, while nonprecious metal (NPM) catalysts are confronted with the challenge of low catalytic activity due to the inverse relationship between the N dissociation barrier and NH ( = 1-3) desorption energy. Herein, we demonstrate NPM (Co, Ni, and Re)-mediated MoCT MXene (where T denotes the OH group) to achieve efficient NH synthesis under mild conditions. In particular, the NH synthesis rate over Re/MoCT and Ni/MoCT can reach 22.

Simple, Efficient, and Scalable Structure-Aware Adapter Boosts Protein Language Models.

Fine-tuning pretrained protein language models (PLMs) has emerged as a prominent strategy for enhancing downstream prediction tasks, often outperforming traditional supervised learning approaches. As a widely applied powerful technique in natural language processing, employing parameter-efficient fine-tuning techniques could potentially enhance the performance of PLMs. However, the direct transfer to life science tasks is nontrivial due to the different training strategies and data forms.