Scientific discovery in the age of artificial intelligence.

Artificial intelligence (AI) is being increasingly integrated into scientific discovery to augment and accelerate research, helping scientists to generate hypotheses, design experiments, collect and interpret large datasets, and gain insights that might not have been possible using traditional scientific methods alone. Here we examine breakthroughs over the past decade that include self-supervised learning, which allows models to be trained on vast amounts of unlabelled data, and geometric deep learning, which leverages knowledge about the structure of scientific data to enhance model accuracy and efficiency. Generative AI methods can create designs, such as small-molecule drugs and proteins, by analysing diverse data modalities, including images and sequences. We discuss how these methods can help scientists throughout the scientific process and the central issues that remain despite such advances. Both developers and users of AI toolsneed a better understanding of when such approaches need improvement, and challenges posed by poor data quality and stewardship remain. These issues cut across scientific disciplines and require developing foundational algorithmic approaches that can contribute to scientific understanding or acquire it autonomously, making them critical areas of focus for AI innovation.

Interface Engineering of the Cu1.5Mn1.5O4/CeO2 Heterostructure for Highly Efficient Electrocatalytic Nitrate Reduction to Ammonia.

The electrochemical nitrate reduction reaction (NO3RR) is considered a sustainable technology to convert the nitrate pollutants to ammonia. However, developing highly efficient electrocatalysts is necessary and challenging given the slow kinetics of the NO3RR with an eight-electron transfer process. Here, a Cu1.5Mn1.5O4 (CMO)/CeO2 heterostructure with rich interfaces is designed and fabricated through an electrospinning and postprocessing technique. Benefiting from the strong coupling between CMO and CeO2, the optimized CMO/CeO2-2 catalyst presents excellent NO3RR performance, with NH3 Faraday efficiency (FE) up to 93.07 ± 1.45% at -0.481 V vs reversible hydrogen electrode (RHE) and NH3 yield rate up to 48.06 ± 1.32 mg cm-2 h-1 at -0.681 V vs RHE. Theoretical calculations demonstrate that the integration of CeO2 with CMO modulates the adsorption/desorption process of the reactants and intermediates, showing a reduced energy barrier in the rate determination step of NO* to N* and achieving an outstanding NO3RR performance.

Single-Image-Based Deep Learning for Precise Atomic Defect Identification.

Defect engineering is widely used to impart the desired functionalities on materials. Despite the widespread application of atomic-resolution scanning transmission electron microscopy (STEM), traditional methods for defect analysis are highly sensitive to random noise and human bias. While deep learning (DL) presents a viable alternative, it requires extensive amounts of training data with labeled ground truth. Herein, employing cycle generative adversarial networks (CycleGAN) and U-Nets, we propose a method based on a single experimental STEM image to tackle high annotation costs and image noise for defect detection. Not only atomic defects but also oxygen dopants in monolayer MoS2 are visualized. The method can be readily extended to other two-dimensional systems, as the training is based on unit-cell-level images. Therefore, our results outline novel ways to train the model with minimal data sets, offering great opportunities to fully exploit the power of DL in the materials science community.

Single-cell transcriptome analysis reveals tumoral microenvironment heterogenicity and hypervascularization in human carotid body tumor.

Carotid body tumor (CBT) is a rare neck tumor located at the adventitia of the common carotid artery bifurcation. The prominent pathological features of CBT are high vascularization and abnormal proliferation. However, single-cell transcriptome analysis of the microenvironment composition and molecular complexity in CBT has yet to be performed. In this study, we performed single-cell RNA sequencing (scRNA-seq) analysis on human CBT to define the cells that contribute to hypervascularization and chronic hyperplasia. Unbiased clustering analysis of transcriptional profiles identified 16 distinct cell populations including endothelial cells (ECs), smooth muscle cells (SMCs), neuron cells, macrophage cells, neutrophil cells, and T cells. Within the ECs population, we defined subsets with angiogenic capacity plus clear signs of later endothelial progenitor cells (EPCs) to normal ECs. Two populations of macrophages were detectable in CBT, macrophage1 showed enrichment in hypoxia-inducible factor-1 (HIF-1) and as well as an early EPCs cell-like population expressing CD14 and vascular endothelial growth factor. In addition to HIF-1-related transcriptional protein expression, macrophages1 also display a neovasculogenesis-promoting phenotype. SMCs included three populations showing platelet-derived growth factor receptor beta and vimentin expression, indicative of a cancer-associated fibroblast phenotype. Finally, we identified three types of neuronal cells, including chief cells and sustentacular cells, and elucidated their distinct roles in the pathogenesis of CBT and abnormal proliferation of tumors. Overall, our study provided the first comprehensive characterization of the transcriptional landscape of CBT at scRNA-seq profiles, providing novel insights into the mechanisms underlying its formation.

Highly diverse sputum microbiota correlates with the disease severity in patients with community-acquired pneumonia: a longitudinal cohort study.

BACKGROUND: Community-acquired pneumonia (CAP) is a common and serious condition that can be caused by a variety of pathogens. However, much remains unknown about how these pathogens interact with the lower respiratory commensals, and whether any correlation exists between the dysbiosis of the lower respiratory microbiota and disease severity and prognosis. METHODS: We conducted a retrospective cohort study to investigate the composition and dynamics of sputum microbiota in patients diagnosed with CAP. In total, 917 sputum specimens were collected consecutively from 350 CAP inpatients enrolled in six hospitals following admission. The V3-V4 region of the 16 S rRNA gene was then sequenced. RESULTS: The sputum microbiota in 71% of the samples were predominately composed of respiratory commensals. Conversely, 15% of the samples demonstrated dominance by five opportunistic pathogens. Additionally, 5% of the samples exhibited sterility, resembling the composition of negative controls. Compared to non-severe CAP patients, severe cases exhibited a more disrupted sputum microbiota, characterized by the highly dominant presence of potential pathogens, greater deviation from a healthy state, more significant alterations during hospitalization, and sparser bacterial interactions. The sputum microbiota on admission demonstrated a moderate prediction of disease severity (AUC = 0.74). Furthermore, different pathogenic infections were associated with specific microbiota alterations. Acinetobacter and Pseudomonas were more abundant in influenza A infections, with Acinetobacter was also enriched in Klebsiella pneumoniae infections. CONCLUSION: Collectively, our study demonstrated that pneumonia may not consistently correlate with severe dysbiosis of the respiratory microbiota. Instead, the degree of microbiota dysbiosis was correlated with disease severity in CAP patients.

Effects of Vibrio harveyi infection on the biochemistry, histology and transcriptome in the hepatopancreas of ivory shell (Babylonia areolata).

The ivory shell (Babylonia areolata) is one of the most promising high quality marine products. However, ivory shell is susceptible to Vibrio harveyi infection during the culture period. In this study, we investigated the biochemical indicators, histological changes and transcriptomic response in the hepatopancreas of ivory shells from the PBS control group (PC) and infection group (A3) with 1 × 109 CFU/mL V. harveyi after 24 h. Results showed that compared to the PC group, biochemical indicators, including malondialdehyde (MDA), reactive oxygen species (ROS), acid phosphatase (ACP), and Caspase 3 (Casp-3) were significantly increased (p < 0.05) in A3 group after V. harveyi infection for 24 h. Compared with the PC group, the hepatopancreas of A3 group were seriously damaged, the columnar epithelial cells of the tissue were enlarged, the space of digestive cells was increased, and vacuolar cavities appeared. A total of 95,581 unigenes were obtained and 2949 (1787 up-regulated and 1162 down-regulated) differential expressed genes (DEGs) were identified in the A3 group. GO and KEGG enrichment analysis showed that DEGs were mainly enriched in immune system process (GO:0002376), antioxidant activity (GO:0016209), lysosome (ko04142), toll and IMD signaling pathway (ko04624), and etc. These biological functions and pathways are associated with immune and inflammatory responses and apoptosis. 12 DEGs were randomly selected for real-time quantitative PCR (RT-qPCR) validation, and the expression profiles of these DEGs were consistent with the transcriptome data, confirming the accuracy and reliability of the transcriptome results. In summary, V. harveyi infection of ivory shells inducing oxidative stress, leading to severe hepatopancreatic damage, stimulating glutathione production to neutralize excessive ROS, and stimulating antimicrobial peptides production to counteract the deleterious effects of bacterial infection, which in turn modifying the immune and inflammatory response, ultimately resulting in apoptosis. This study provided valuable information to explore the immune regulation mechanism after V. harveyi infection and established molecular basis to support the prevention of V. harveyi infection.

Interactions between inhalable aged microplastics and lung surfactant: Potential pulmonary health risks.

The relationship between microplastics (MPs) and human respiratory health has garnered significant attention since inhalation constitutes the primary pathway for atmospheric MP exposure. While recent studies have revealed respiratory risks associated with MPs, virgin MPs used as plastic surrogates in these experiments did not represent the MPs that occur naturally and that undergo aging effects. Thus, the effects of aged MPs on respiratory health remain unknown. We herein analyzed the interaction between inhalable aged MPs with lung surfactant (LS) extracted from porcine lungs vis-à-vis interfacial chemistry employing in-vitro experiments, and explored oxidative damage induced by aged MPs in simulated lung fluid (SLF) and the underlying mechanisms of action. Our results showed that aged MPs significantly increased the surface tension of the LS, accompanied by a diminution in its foaming ability. The stronger adsorptive capacity of the aged MPs toward the phospholipids of LS appeared to produce increased surface tension, while the change in foaming ability might have resulted from a variation in the protein secondary structure and the adsorption of proteins onto MPs. The adsorption of phospholipid and protein components then led to the aggregation of MPs in SLF, where the aged MPs exhibited smaller hydrodynamic diameters in comparison with the unaged MPs, likely interacting with biomolecules in bodily fluids to exacerbate health hazards. Persistent free radicals were also formed on aged MPs, inducing the formation of reactive oxygen species such as superoxide radicals (O2•-), hydrogen peroxide (HOOH), and hydroxyl radicals (•OH); this would lead to LS lipid peroxidation and protein damage and increase the risk of respiratory disease. Our investigation was the first-ever to reveal a potential toxic effect of aged MPs and their actions on the human respiratory system, of great significance in understanding the risk of inhaled MPs on lung health.

Revealing the molecular structures of α-Al2O3(0001)-water interface by machine learning based computational vibrational spectroscopy.

Solid-water interfaces are crucial to many physical and chemical processes and are extensively studied using surface-specific sum-frequency generation (SFG) spectroscopy. To establish clear correlations between specific spectral signatures and distinct interfacial water structures, theoretical calculations using molecular dynamics (MD) simulations are required. These MD simulations typically need relatively long trajectories (a few nanoseconds) to achieve reliable SFG response function calculations via the dipole moment-polarizability time correlation function. However, the requirement for long trajectories limits the use of computationally expensive techniques, such as ab initio MD (AIMD) simulations, particularly for complex solid-water interfaces. In this work, we present a pathway for calculating vibrational spectra (IR, Raman, and SFG) of solid-water interfaces using machine learning (ML)-accelerated methods. We employ both the dipole moment-polarizability correlation function and the surface-specific velocity-velocity correlation function approaches to calculate SFG spectra. Our results demonstrate the successful acceleration of AIMD simulations and the calculation of SFG spectra using ML methods. This advancement provides an opportunity to calculate SFG spectra for complicated solid-water systems more rapidly and at a lower computational cost with the aid of ML.

Acoustic phonon excitation in gold probed by time-resolved photoemission electron microscopy.

Electron-phonon coupling is an important energy transfer mechanism in solids after ultrafast laser excitation. In this study, we present an extreme ultraviolet (EUV) and infrared (IR) pump-probe photoemission experiment to investigate the electron-phonon coupling in nonequilibrium gold. The energy of IR-laser-emitted photoelectrons is shifted due to the EUV photoemission and oscillates with a ∼4THz frequency. Such oscillation is considered as the effective excitation of the longitudinal acoustic phonon mode in gold through the spectral-dependent electron-phonon coupling. Our study showcases the capability of time-resolved photoemission electron microscopy to monitor the non-equilibrium lattice vibrations with ultrahigh spatial and temporal resolution.

Imaging and Controlling Ultrafast Electron Pulses Emitted from Plasmonic Nanostructures.

We experimentally study photoemission from gold nanodisk arrays using space-, time-, and energy-resolved photoemission electron microscopy. When excited by a plasmonic resonant infrared (IR) laser pulse, plasmonic hotspots are generated owing to local surface plasmon resonance. Photoelectrons emitted from each plasmonic hotspot form a nanoscale and ultrashort electron pulse. When the system is excited by an extreme ultraviolet (EUV) laser pulse, a uniformly distributed photoelectron cloud is formed across the sample surface. When excited by the IR and EUV laser pulses together, both the photoemission image and kinetic energy vary significantly for the IR laser-generated electrons depending on the time delay between the two laser pulses. These observations are well explained by the Coulomb interaction with the EUV laser-generated electron cloud. Our study offers a feasible approach to manipulate the energy of electron pulse emitted from a plasmonic nanostructure on an ultrafast time scale.

MSC Exosomes Containing Valproic Acid Promote Wound Healing by Modulating Inflammation and Angiogenesis.

PURPOSE: Chronic wounds that are difficult to heal pose a major challenge for clinicians and researchers. Currently, common treatment methods focus on isolating the wound from the outside world, relying on the tissue at the wound site to grow and heal unaided. Umbilical cord mesenchymal stem cell (MSC) exosomes can promote wound healing by enhancing new blood vessel growth at the wound site. Valproic acid (VPA) reduces the inflammatory response and acts on macrophages to accelerate wound closure. In this study, VPA was loaded into umbilical cord MSC exosomes to form a drug carrier exosome (VPA-EXO) with the aim of investigating the effect of VPA-EXO on wound healing. METHODS: This study first isolated and obtained umbilical cord MSC exosomes, then added VPA to the exosomes and explored the ability of VPA-EXO to promote the proliferation and migration of human skin fibroblasts (HSFs) and human umbilical vein endothelial cells (HUVECs), as well as the ability to promote the angiogenesis of HUVECs, by using scratch, Transwell, and angiogenesis assays. An in vitro cell model was established and treated with VPA-EXO, and the expression levels of inflammation and pro-angiogenesis-related proteins and genes were examined using Western blot and qRT-PCR. The therapeutic effect of VPA-EXO on promoting wound healing in a whole skin wound model was investigated using image analysis of the wound site, H&E staining, and immunohistochemical staining experiments in a mouse wound model. RESULTS: The in vitro model showed that VPA-EXO effectively promoted the proliferation and migration of human skin fibroblast cells and human umbilical vein endothelial cells; significantly inhibited the expression of MMP-9, IL-1ß, IL-8, TNF-α, and PG-E2; and promoted the expression of vascular endothelial growth factors. In the mouse wound model, VPA-EXO reduced inflammation at the wound site, accelerated wound healing, and significantly increased the collagen content of tissue at the wound site. CONCLUSIONS: As a complex with dual efficacy in simultaneously promoting tissue regeneration and inhibiting inflammation, VPA-EXO has potential applications in tissue wound healing and vascular regeneration. In future studies, we will further investigate the mechanism of action and application scenarios of drug-loaded exosome complexes in different types of wound healing and vascular regeneration.

Workplace-based primary prevention intervention reduces incidence of hypertension: a post hoc analysis of cluster randomized controlled study.

BACKGROUND: A workplace-based primary prevention intervention be an effective approach to reducing the incidence of hypertension (HTN). However, few studies to date have addressed the effect among the Chinese working population. We assessed the effect of a workplace-based multicomponent prevention interventions program for cardiovascular disease on reducing the occurrence of HTN through encouraging employees to adopt a healthy lifestyle. METHODS: In this post hoc analysis of cluster randomized controlled study, 60 workplaces across 20 urban regions in China were randomized to either the intervention group (n = 40) or control group (n = 20). All employees in each workplace were asked to complete a baseline survey after randomization for obtaining sociodemographic information, health status, lifestyle, etc. Employees in the intervention group were given a 2-year workplace-based primary prevention intervention program for improving their cardiovascular health, including (1) cardiovascular health education, (2) a reasonable diet, (3) tobacco cessation, (4) physical environment promotion, (5) physical activity, (6) stress management, and (7) health screening. The primary outcome was the incidence of HTN, and the secondary outcomes were improvements of blood pressure (BP) levels and lifestyle factors from baseline to 24 months. A mix effect model was used to assess the intervention effect at the end of the intervention in the two groups. RESULTS: Overall, 24,396 participants (18,170 in the intervention group and 6,226 in the control group) were included (mean [standard deviation] age, 39.3 [9.1] years; 14,727 men [60.4%]). After 24 months of the intervention, the incidence of HTN was 8.0% in the intervention groups and 9.6% in the control groups [relative risk (RR) = 0.66, 95% CI, 0.58 ~ 0.76, P < 0.001]. The intervention effect was significant on systolic BP (SBP) level (ß = - 0.7 mm Hg, 95% CI, - 1.06 ~ - 0.35; P < 0.001) and on diastolic BP (DBP) level (ß = - 1.0 mm Hg, 95% CI, - 1.31 ~ - 0.76; P < 0.001). Moreover, greater improvements were reported in the rates of regular exercise [odd ratio (OR) = 1.39, 95% CI, 1.28 ~ 1.50; P < 0.001], excessive intake of fatty food (OR = 0.54, 95% CI, 0.50 ~ 0.59; P < 0.001), and restrictive use of salt (OR = 1.22, 95% CI, 1.09 ~ 1.36; P = 0.001) in intervention groups. People with a deteriorating lifestyle had higher rates of developing HTN than those with the same or improved lifestyle. Subgroup analysis showed that the intervention effect of BP on employees with educational attainment of high school above (SBP: ß = - 1.38/ - 0.76 mm Hg, P < 0.05; DBP: ß = - 2.26/ - 0.75 mm Hg, P < 0.001), manual labor workers and administrative worker (SBP: ß = - 1.04/ - 1.66 mm Hg, P < 0.05; DBP: ß = - 1.85/ - 0.40 mm Hg, P < 0.05), and employees from a workplace with an affiliated hospital (SBP: ß = - 2.63 mm Hg, P < 0.001; DBP: ß = - 1.93 mm Hg, P < 0.001) were significantly in the intervention group. CONCLUSIONS: This post hoc analysis found that workplace-based primary prevention interventions program for cardiovascular disease were effective in promoting healthy lifestyle and reducing the incidence of HTN among employees. TRIAL REGISTRATION: Chinese Clinical Trial Registry No. ChiCTR-ECS-14004641.

Trajectory-controlled high-order harmonic generation in ZnO crystals.

We experimentally and theoretically study high-order harmonic generation in zinc oxide crystals irradiated by mid-infrared lasers. The trajectories are mapped to the far field spatial distribution of harmonics. The divergence angles of on-axis and off-axis parts exhibit different dependences on the order of the harmonics. This observation can be theoretically reproduced by the coherent interference between the short and long trajectories with dephasing time longer than 0.5 optical cycle. Further, the relative contribution of the short and long trajectories is demonstrated to be accurately controlled by a one-color or two-color laser on the attosecond time scale. This work provides a reliable method to determine the electron dephasing time and demonstrates a versatile control of trajectory interference in the solid high-order harmonic generation.

Deep Variational Free Energy Approach to Dense Hydrogen.

We developed a deep generative model-based variational free energy approach to the equations of state of dense hydrogen. We employ a normalizing flow network to model the proton Boltzmann distribution and a fermionic neural network to model the electron wave function at given proton positions. By jointly optimizing the two neural networks we reached a comparable variational free energy to the previous coupled electron-ion Monte Carlo calculation. The predicted equation of state of dense hydrogen under planetary conditions is denser than the findings of ab initio molecular dynamics calculation and empirical chemical model. Moreover, direct access to the entropy and free energy of dense hydrogen opens new opportunities in planetary modeling and high-pressure physics research.

Entangled X-Ray Photon Pair Generation by Free-Electron Lasers.

We investigate entangled x-ray photon pair emissions in a free-electron laser (FEL) and establish a quantum electrodynamical theory for coherently amplified entangled photon pair emission from microbunched electron pulses in the undulator. We provide a scheme to generate highly entangled x-ray photon pairs and numerically demonstrate the properties of entangled emission, which is of great importance in x-ray quantum optics. Our work shows a unique advantage of FELs in entangled x-ray photon pair generation.

Localized Elevation of Wall Shear Stress Is Linked to Recent Symptoms in Patients with Carotid Stenosis.

BACKGROUND: Estimation of symptomatic and asymptomatic carotid atherosclerosis differences can be the basis for prevention and management of carotid artery stenosis disease. In clinical practice, carotid plaque vulnerability is assessed only on the basis of luminal stenosis. However, the evolution of carotid plaque from an asymptomatic state to a symptomatic one is a complex process and the underlying hemodynamic mechanisms are unknown. We aimed to investigate the differences in hemodynamic parameters between patients with recently symptomatic carotid stenosis and asymptomatic ones. METHODS: Hemodynamic simulations were performed on 26 carotid plaques from 25 patients with carotid artery stenosis ≥50%, 16 of whom had recent cerebrovascular ischemic events. Using human-specific flow parameters and 3D reconstruction of carotid computed tomography angiography images, we assessed hemodynamic characteristics such as wall shear stress (WSS), time-averaged WSS (TAWSS), oscillatory shear index, and relative residence time (RRT) during the cardiac cycle in patients with and without symptoms. RESULTS: We found that symptomatic carotid stenosis patients had greater local TAWSS (99.59 ± 26.29 vs. 60.40 ± 20.46 dyn/cm2, p = 0.0007) and maximal WSS (116.65 ± 39.11 vs. 68.28 ± 23.67 dyn/cm2, p = 0.003), but lower RRT (0.019 ± 0.006 vs. 0.013 ± 0.069 s, p = 0.049), than asymptomatic patients, but this hemodynamic difference was not associated with carotid stenosis severity (p = 0.70). Patients with transient ischemic attack (TIA) or stroke had higher local TAWSS and WSSmax than patients with asymptomatic stenosis (p < 0.05). Subgroup analysis showed that there was no statistical difference in local hemodynamic variables between TIA and stroke patients with carotid artery stenosis. CONCLUSIONS: Patients with carotid artery stenosis are more likely to experience acute ischemic cerebrovascular accidents if they have higher WSS. Simultaneous assessment with hemodynamic parameters like WSS along with stenosis severity may aid risk stratification in patients with asymptomatic carotid artery stenosis.

Geometric Phase Effect in Attosecond Stimulated X-ray Raman Spectroscopy.

Conical intersections (CIs) are diabolical points in the potential energy surfaces generally caused by point-wise degeneracy of different electronic states, and give rise to the geometric phases (GPs) of molecular wave functions. Here we theoretically propose and demonstrate that the transient redistribution of ultrafast electronic coherence in attosecond Raman signal (TRUECARS) spectroscopy is capable of detecting the GP effect in excited state molecules by applying two probe pulses including an attosecond and a femtosecond X-ray pulse. The mechanism is based on a set of symmetry selection rules in the presence of nontrivial GPs. The model of this work can be realized for probing the geometric phase effect in the excited state dynamics of complex molecules with appropriate symmetries, using attosecond light sources such as free-electron X-ray lasers.

Machine learning assisted coarse-grained molecular dynamics modeling of meso-scale interfacial fluids.

A hallmark of meso-scale interfacial fluids is the multi-faceted, scale-dependent interfacial energy, which often manifests different characteristics across the molecular and continuum scale. The multi-scale nature imposes a challenge to construct reliable coarse-grained (CG) models, where the CG potential function needs to faithfully encode the many-body interactions arising from the unresolved atomistic interactions and account for the heterogeneous density distributions across the interface. We construct the CG models of both single- and two-component polymeric fluid systems based on the recently developed deep coarse-grained potential [Zhang et al., J. Chem. Phys. 149, 034101 (2018)] scheme, where each polymer molecule is modeled as a CG particle. By only using the training samples of the instantaneous force under the thermal equilibrium state, the constructed CG models can accurately reproduce both the probability density function of the void formation in bulk and the spectrum of the capillary wave across the fluid interface. More importantly, the CG models accurately predict the volume-to-area scaling transition for the apolar solvation energy, illustrating the effectiveness to probe the meso-scale collective behaviors encoded with molecular-level fidelity.

DeePMD-kit v2: A software package for deep potential models.

DeePMD-kit is a powerful open-source software package that facilitates molecular dynamics simulations using machine learning potentials known as Deep Potential (DP) models. This package, which was released in 2017, has been widely used in the fields of physics, chemistry, biology, and material science for studying atomistic systems. The current version of DeePMD-kit offers numerous advanced features, such as DeepPot-SE, attention-based and hybrid descriptors, the ability to fit tensile properties, type embedding, model deviation, DP-range correction, DP long range, graphics processing unit support for customized operators, model compression, non-von Neumann molecular dynamics, and improved usability, including documentation, compiled binary packages, graphical user interfaces, and application programming interfaces. This article presents an overview of the current major version of the DeePMD-kit package, highlighting its features and technical details. Additionally, this article presents a comprehensive procedure for conducting molecular dynamics as a representative application, benchmarks the accuracy and efficiency of different models, and discusses ongoing developments.