Search

Abstract Reliable molecular property prediction is essential for various scientific endeavors and industrial applications, such as drug discovery. However, the data scarcity, combined with the highly non-linear causal relationships between physicochemical and biological properties and conventional molecular featurization schemes, complicates the development of robust molecular machine learning models. Self-supervised learning (SSL) has emerged as a popular solution, utilizing large-scale, unannotated molecular data to learn a foundational representation of chemical space that might be advantageous for downstream tasks. Yet, existing molecular SSL methods largely overlook chemical knowledge, including molecular structure similarity, scaffold composition, and the context-dependent aspects of molecular properties when operating over the chemical space. They also struggle to learn the subtle variations in structure-activity relationship. This paper introduces a multi-channel pre-training framework that learns robust and generalizable chemical knowledge. It leverages the structural hierarchy within the molecule, embeds them through distinct pre-training tasks across channels, and aggregates channel information in a task-specific manner during fine-tuning. Our approach demonstrates competitive performance across various molecular property benchmarks and offers strong advantages in particularly challenging yet ubiquitous scenarios like activity cliffs.

Geospatial Artificial Intelligence (GeoAI), as the integration of geospatial studies and AI, has become one of the fastest-developing research directions in spatial data science and geography. This rapid change in the field calls for a deeper understanding of the recent developments and envision where the field is going in the near future. In this work, we provide a quantitative analysis of the GeoAI literature from the spatial, temporal, and semantic aspects. We briefly discuss the history of AI and GeoAI by highlighting some pioneering work. Then we discuss the current landscape of GeoAI by selecting five representative subdomains including remote sensing, urban computing, Earth system science, cartography, and geospatial semantics. Finally, we highlight several unique future research directions of GeoAI which are classified into two groups: GeoAI method development challenges and GeoAI Ethics challenges. Topics include heterogeneity-aware GeoAI, knowledge-guided GeoAI, spatial representation learning, geo-foundation models, fairness-aware GeoAI, privacy-aware GeoAI, as well as interpretable and explainable GeoAI. We hope our review of GeoAI’s past, present, and future is comprehensive and can enlighten the next generation of GeoAI research.

Prediction of dynamic environmental variables in unmonitored sites remains a long-standing challenge for water resources science. The majority of the world’s freshwater resources have inadequate monitoring of critical environmental variables needed for management. Yet, the need to have widespread predictions of hydrological variables such as river flow and water quality has become increasingly urgent due to climate and land use change over the past decades, and their associated impacts on water resources. Modern machine learning methods increasingly outperform their process-based and empirical model counterparts for hydrologic time series prediction with their ability to extract information from large, diverse data sets. We review relevant state-of-the art applications of machine learning for streamflow, water quality, and other water resources prediction and discuss opportunities to improve the use of machine learning with emerging methods for incorporating watershed characteristics and process knowledge into classical, deep learning, and transfer learning methodologies. The analysis here suggests most prior efforts have been focused on deep learning frameworks built on many sites for predictions at daily time scales in the United States, but that comparisons between different classes of machine learning methods are few and inadequate. We identify several open questions for time series predictions in unmonitored sites that include incorporating dynamic inputs and site characteristics, mechanistic understanding and spatial context, and explainable AI techniques in modern machine learning frameworks.

Abstract Superalloy Inconel718 is an important material for aircraft preparation because of its excellent performance at high temperatures. However, when cutting Inconel718, a large amount of cutting heat will be generated, resulting in excessive tool temperature and serious wear, which accelerates the tool failure. In order to solve this problem, the influence of tool angle on the process of thermal aided machining was studied by simulation model combined with thermal aided machining technology. During the cutting process, the workpiece preheating temperature rises from room temperature 20° C to 500° C, the front tool angle range is − 5° to 10°, and the rear tool angle range is 4° to 16°. By analyzing various parameters, it was found that a smaller tool rake angle can effectively reduce the tool temperature. Additionally, a flank angle of around 12° was found to decrease the maximum wear area of the tool by approximately 10.5%. Moreover, it was observed that implementing heat-assisted machining can result in a significant reduction of tool temperature by 11.1%, as well as a decrease in cutting force ranging from 18 to 22%, particularly at temperatures exceeding 500 °C.

Abstract Accurate and cost-effective quantification of the carbon cycle for agroecosystems at decision-relevant scales is critical to mitigating climate change and ensuring sustainable food production. However, conventional process-based or data-driven modeling approaches alone have large prediction uncertainties due to the complex biogeochemical processes to model and the lack of observations to constrain many key state and flux variables. Here we propose a Knowledge-Guided Machine Learning (KGML) framework that addresses the above challenges by integrating knowledge embedded in a process-based model, high-resolution remote sensing observations, and machine learning (ML) techniques. Using the U.S. Corn Belt as a testbed, we demonstrate that KGML can outperform conventional process-based and black-box ML models in quantifying carbon cycle dynamics. Our high-resolution approach quantitatively reveals 86% more spatial detail of soil organic carbon changes than conventional coarse-resolution approaches. Moreover, we outline a protocol for improving KGML via various paths, which can be generalized to develop hybrid models to better predict complex earth system dynamics.

The coal-based energy consumption structure will significantly increase carbon emissions. Along with the rise in per-capita income, this energy consumption structure will reduce economy sustainable growth. The measure of sustainable economic growth is to use the DEA-SBM super efficiency model of unexpected output and calculate the green total factor productivity (GTFP). The proportion of coal consumption in energy consumption represents the structure of energy consumption . The panel data in this paper are from the records of various provinces and cities in China from 2005 to 2017 (except Tibet). In the results of the mixed panel regression, the impact of energy consumption structure on sustainable economic growth is not significantly positive, and after controlling for time and provinces, the overall impact of energy consumption structure on sustainable economic growth is positive, and significant at the 5% level, and its impact on carbon dioxide emissions is significant at the 1% level. It means that the coal-based energy consumption structure is a significant factor in increasing carbon emissions. Due to the heterogeneity of various provinces, the energy consumption structure of developed provinces and cities will significantly reduce sustainable economic growth at the 1% level, while it will increase significantly at the 5% level in less developed provinces and cities, which has uncertain impact on moderately developed provinces and cities. For developed provinces, reducing the proportion of coal consumption, improving the energy consumption structure can bring about the improvement of GTFP, that is, to promote sustainable economic growth. For other provinces, strategies to change the structure of energy consumption should be put forward according to the stage of economic development. Using the mediating effect model, it is found that both developed and underdeveloped provinces contribute to sustainable economic growth through carbon dioxide emissions. Therefore, controlling carbon dioxide emissions is very important for sustainable economic growth.

Deep learning (DL) models are increasingly used to forecast water quality variables for use in decision making. Ingesting recent observations of the forecasted variable has been shown to greatly increase model performance at monitored locations; however, observations are not collected at all locations, and methods are not yet well developed for DL models for optimally ingesting recent observations from other sites to inform focal sites. In this paper, we evaluate two different DL model structures, a long short-term memory neural network (LSTM) and a recurrent graph convolutional neural network (RGCN), both with and without data assimilation for forecasting daily maximum stream temperature 7 days into the future at monitored and unmonitored locations in a 70-segment stream network. All our DL models performed well when forecasting stream temperature as the root mean squared error (RMSE) across all models ranged from 2.03 to 2.11°C for 1-day lead times in the validation period, with substantially better performance at gaged locations (RMSE = 1.45–1.52°C) compared to ungaged locations (RMSE = 3.18–3.27°C). Forecast uncertainty characterization was near-perfect for gaged locations but all DL models were overconfident (i.e., uncertainty bounds too narrow) for ungaged locations. Our results show that the RGCN with data assimilation performed best for ungaged locations and especially at higher temperatures (>18°C) which is important for management decisions in our study location. This indicates that the networked model structure and data assimilation techniques may help borrow information from nearby monitored sites to improve forecasts at unmonitored locations. Results from this study can help guide DL modeling decisions when forecasting other important environmental variables.

We conducted an 8-week feeding trial to explore the effects of dietary soybean lecithin (SBL) on growth performance and metabolic abilities in juvenile largemouth bass Micropterus salmoides. The addition of 4 % SBL to the feed significantly increases the specific growth rate (SGR), carcass ratio (CR) and crude protein in the fish body and muscle compared with control groups (0 %) (P 0.05). In addition, 4 % SBL elevated activities and mRNA levels of FAS in the fish liver compared with controls (P

Accurate climate data at fine spatial resolution are essential for scientific research and the development and planning of crucial social systems, such as energy and agriculture. Among them, sea surface temperature plays a critical role as the associated El Niño–Southern Oscillation (ENSO) is considered a significant signal of the global interannual climate system. In this paper, we propose an implicit neural representation-based interpolation method with temporal information (T_INRI) to reconstruct climate data of high spatial resolution, with sea surface temperature as the research object. Traditional deep learning models for generating high-resolution climate data are only applicable to fixed-resolution enhancement scales. In contrast, the proposed T_INRI method is not limited to the enhancement scale provided during the training process and its results indicate that it can enhance low-resolution input by arbitrary scale. Additionally, we discuss the impact of temporal information on the generation of high-resolution climate data, specifically, the influence of the month from which the low-resolution sea surface temperature data are obtained. Our experimental results indicate that T_INRI is advantageous over traditional interpolation methods under different enhancement scales, and the temporal information can improve T_INRI performance for a different calendar month. We also examined the potential capability of T_INRI in recovering missing grid value. These results demonstrate that the proposed T_INRI is a promising method for generating high-resolution climate data and has significant implications for climate research and related applications.

A total of 99 endophytic fungal isolates were obtained from the roots of Glycyrrhiza inflata, which was a traditional medicinal plant mainly distributed in Xinjiang, China. Twenty-two distinct isolates were selected for further taxonomical identification by morphological traits and internal transcribed spacer (ITS) rRNA gene sequence analysis. Eleven genera were identified, among which Aspergillus, Alternaria and Fusarium were dominant. The crude extracts of 22 distinct identified fungi were successively evaluated for their antifungal activities on three rice fungal pathogens using the method of hyphal radial growth rate. Among them, the crude extract of Alternaria an-gustiovoidea Glinf007 showed the significantly mycelial growth inhibitory activity. The results demonstrated that G. inflata contained a diversity of culturable endophytic fungi, which could produce natural antimicrobial compounds that might be of great value to the agriculture and pharmaceutical industries.

An 8-week feeding study was conducted to examine how dietary pantothenic acid (PA) affects growth, antioxidant ability and innate immunity in juvenile black carp (Mylopharyngodon piceus). Juvenile fish (7.35 ± 0.06 g) were fed diets with PA at 0, 5, 10, 20, 40 and 80 mg/kg, respectively. The highest weight gain (WG), specific growth rate (SGR) and protein efficiency ratio (PER) were obtained with 10 mg/kg dietary PA (p

Immune cell infiltration into the tumor microenvironment is associated with cancer prognosis. Tumor-associated macrophages play essential roles in tumor initiation, progression, and metastasis. Follistatin-like protein 1 (FSTL1), a widely expressed glycoprotein in human and mouse tissues, is a tumor suppressor in various cancers and a regulator of macrophage polarization. However, the mechanism by which FSTL1 affects crosstalk between breast cancer cells and macrophages remains unclear. By analyzing public data, we found that FSTL1 expression was significantly low in breast cancer tissues compared to normal breast tissues, and high expression of FSTL1 in patients indicated prolonged survival. Using flow cytometry, we found that total and M2-like macrophages dramatically increased in the metastatic lung tissues during breast cancer lung metastasis in Fstl1+/− mice. Transwell assay in vitro and q-PCR experimental results showed that FSTL1 inhibited macrophage migration toward 4T1 cells by decreasing CSF1, VEGF-α, and TGF-β secretion in 4T1 cells. We demonstrated that FSTL1 inhibited M2-like tumor-associated macrophage recruitment toward the lungs by suppressing CSF1, VEGF-α, and TGF-β secretion in 4T1 cells. Therefore, we identified a potential therapeutic strategy for triple-negative breast cancer.

This study evaluated the effects of dietary soybean lecithin (SBL) on the growth, haematological indices, immunities, antioxidant capabilities, and inflammatory and intestinal barrier functions because little information of dietary SBL could be obtained in juvenile largemouth bass (Micropterus salmoides). The fish were fed identical diets except for SBL added at 0, 2, 4 and 8%. It was found that 4 and 8% SBL significantly increased fish weight gain and daily growth rate (p < 0.05), while 4% SBL was optimal for enhancing RBC, HGB, PLT, MCV, MCH, WBC and MON in blood, and ALB and ALP in serum (p < 0.05). SBL (4%) also significantly elevated the antioxidant enzymes activities of T-SOD, CAT, GR, GPx, GST and T-AOC and GSH contents; increased mRNA transcription levels of Nrf2, Cu/Zn-SOD, CAT, GR, GST3 and GPx3; and decreased MDA contents. Keap1a and Keap1b levels were markedly down-regulated (p < 0.05). SBL (4%) significantly enhanced levels of the immune factors (ACP, LZM and C3) and the mRNA expression levels of innate immune-related genes (C3, C4, CFD, HEPC and MHC-I) compared with the control groups (0%) (p < 0.05). SBL (4%) significantly increased IgM and T-NOS in the intestine (p < 0.05) and significantly decreased levels of TNF-α, IL-8, IL-1β and IFN-γ and increased TGF-β1 at both transcription and protein levels in the liver and intestine (p < 0.05). The mRNA expression levels of MAPK13, MAPK14 and NF-κB P65 were significantly decreased in the intestine in the 4% SBL groups (p < 0.05). Histological sections also demonstrated that 4% SBL protected intestinal morphological structures compared with controls. This included increased intestinal villus height and muscular thickness (p < 0.05). Furthermore, the mRNA expression levels of the intestinal epithelial cell tight junction proteins (TJs) (ZO-1, claudin-3, claudin-4, claudin-5, claudin-23 and claudin-34) and mucin-5AC were significantly up-regulated in the 4% SBL groups compared with the controls (p < 0.05). In conclusion, these results suggested that 4% dietary SBL could not only improve growth, haematological indices, antioxidant capabilities, immune responses and intestinal functions, but also alleviate inflammatory responses, thereby providing reference information for the feed formulations in cultured largemouth bass.

Objective: Although Follistatin-like protein 1 (FSTL1), as an “adipokine”, is highly expressed in preadipocytes, the detail role of FSTL1 in adipogenesis and obesity remains not fully understood. Methods: In vitro differentiation of both Fstl1−/− murine embryonic fibroblasts (MEFs) and stromal vascular fraction (SVF) were measured to assess the specific role of FSTL1 in adipose differentiation. Fstl1 adipocyte-specific knockout mice were generated to evaluate its role in obesity development. Gene expression analysis and phosphorylation patterns were performed to check out the molecular mechanism of the biological function of FSTL1. Results: FSTL1 deficiency inhibited preadipocytes differentiation in vitro and obesity development in vivo. Glycosylation at N142 site was pivotal for the biological effect of FSTL1 during adipogenesis; the conversion between PPARγ and p-PPARγ was the key factor for the function of FSTL1. Molecular mechanism studies showed that FSTL1 functions through the integrin/FAK/ERK signaling pathway. Conclusions: Our results suggest that FSTL1 promotes adipogenesis by inhibiting the conversion of PPARγ to p-PPARγ through the integrin/FAK/ERK signaling pathway. Glycosylated modification at N142 of FSTL1 is the key site to exert its biological effect.

The inositol phospholipid signaling system mediates plant growth, development, and responses to adverse conditions. Diacylglycerol kinase (DGK) is one of the key enzymes in the phosphoinositide-cycle (PI-cycle), which catalyzes the phosphorylation of diacylglycerol (DAG) to form phosphatidic acid (PA). To date, comprehensive genomic and functional analyses of DGKs have not been reported in wheat. In this study, 24 DGK gene family members from the wheat genome (TaDGKs) were identified and analyzed. Each putative protein was found to consist of a DGK catalytic domain and an accessory domain. The analyses of phylogenetic and gene structure analyses revealed that each TaDGK gene could be grouped into clusters I, II, or III. In each phylogenetic subgroup, the TaDGKs demonstrated high conservation of functional domains, for example, of gene structure and amino acid sequences. Four coding sequences were then cloned from Chinese Spring wheat. Expression analysis of these four genes revealed that each had a unique spatial and developmental expression pattern, indicating their functional diversification across wheat growth and development processes. Additionally, TaDGKs were also prominently up-regulated under salt and drought stresses, suggesting their possible roles in dealing with adverse environmental conditions. Further cis-regulatory elements analysis elucidated transcriptional regulation and potential biological functions. These results provide valuable information for understanding the putative functions of DGKs in wheat and support deeper functional analysis of this pivotal gene family. The 24 TaDGKs identified and analyzed in this study provide a strong foundation for further exploration of the biological function and regulatory mechanisms of TaDGKs in response to environmental stimuli.

Monitoring the extent of plateau forests has drawn much attention from governments given the fact that the plateau forests play a key role in global carbon circulation. Despite the recent advances in the remote-sensing applications of satellite imagery over large regions, accurate mapping of plateau forest remains challenging due to limited ground truth information and high uncertainties in their spatial distribution. In this paper, we aim to generate a better segmentation map for plateau forests using high-resolution satellite imagery with limited ground-truth data. We present the first 2 m spatial resolution large-scale plateau forest dataset of Sanjiangyuan National Nature Reserve, including 38,708 plateau forest imagery samples and 1187 handmade accurate plateau forest ground truth masks. We then propose an few-shot learning method for mapping plateau forests. The proposed method is conducted in two stages, including unsupervised feature extraction by leveraging domain knowledge, and model fine-tuning using limited ground truth data. The proposed few-shot learning method reached an F1-score of 84.23%, and outperformed the state-of-the-art object segmentation methods. The result proves the proposed few-shot learning model could help large-scale plateau forest monitoring. The dataset proposed in this paper will soon be available online for the public.

Plantation mapping is important for understanding deforestation and climate change. While most existing plantation products are created manually, in this paper we study an ensemble learning based framework for automatically mapping plantations in southern Kalimantan on a yearly scale using remote sensing data. We study the effectiveness of several components in this framework, including class aggregation, data sampling, learning model selection and post-processing, by comparing with multiple baselines. In addition, we analyze the quality of our plantation mapping product by visual examination of high resolution images. We also compare our method to existing manually labeled plantation datasets and show that our method can achieve a better balance of precision (i.e., user's accuracy) and recall (i.e., producer's accuracy).

River system is critical for the future sustainability of our planet but is always under the pressure of food, water and energy demands. Recent advances in machine learning bring a great potential for automatic river mapping using satellite imagery. Surface river mapping can provide accurate and timely water extent information that is highly valuable for solid policy and management decisions. However, accurate large-scale river mapping remains challenging given limited labels, spatial heterogeneity and noise in satellite imagery (e.g., clouds and aerosols). In this paper, we propose a new multi-source data-driven method for large-scale river mapping by combining multi-spectral imagery and synthetic aperture radar data. In particular, we build a multi-source data segmentation model, which uses contrastive learning to extract the common information between multiple data sources while also preserving distinct knowledge from each data source. Moreover, we create the first large-scale multi-source river imagery dataset based on Sentinel-1 and Sentinel-2 satellite data, along with 1013 handmade accurate river segmentation mask (which will be released to the public). In this dataset, our method has been shown to produce superior performance (F1-score is 91.53%) over multiple state-of-the-art segmentation algorithms. We also demonstrate the effectiveness of the proposed contrastive learning model in mapping river extent when we have limited and noisy data.

Fungal phytotoxic secondary metabolites are poisonous substances to plants produced by fungi through naturally occurring biochemical reactions. These metabolites exhibit a high level of diversity in their properties, such as structures, phytotoxic activities, and modes of toxicity. They are mainly isolated from phytopathogenic fungal species in the genera of Alternaria, Botrytis, Colletotrichum, Fusarium, Helminthosporium, and Phoma. Phytotoxins are either host specific or non-host specific phytotoxins. Up to now, at least 545 fungal phytotoxic secondary metabolites, including 207 polyketides, 46 phenols and phenolic acids, 135 terpenoids, 146 nitrogen-containing metabolites, and 11 others, have been reported. Among them, aromatic polyketides and sesquiterpenoids are the main phytotoxic compounds. This review summarizes their chemical structures, sources, and phytotoxic activities. We also discuss their phytotoxic mechanisms and structure–activity relationships to lay the foundation for the future development and application of these promising metabolites as herbicides.

The Flame pyrolysis method used to synthesize carbon nanotubes was studied in this work. In order to improve the quality of synthesized carbon nanotubes, it is important to change the corresponding natures of the catalyst. Two catalyst preparation methods, namely, the sol-gel method and the impregnation method, were compared in this experiment. The properties of the catalyst are analyzed in depth by energy dispersive spectrometer (EDS), x-ray diffraction (XRD), temperature program reduction (TPR). The generation of carbon nanotubes was systematically analysed through scanning electron microscope (SEM), molecule dynamics (MD), raman spectroscopy and transmission electron microscope (TEM). The results show that the catalysts prepared by the impregnation method are stickier, dispersed and easier to dip onto the probe or substrate, which is beneficial for the large-scale production of carbon tubes. The specific surface area of alumina is larger and the iron and molybdenum oxide are more evenly dispersed on the surface of alumina. The carbon nanotubes produced by the catalysts prepared by impregnation method are flatter and have less impurities. The ratio of ID/IG+ is 29.7% lower than that of the sol-gel method in the Raman spectra. The TEM statistics show that the average diameter of the carbon tubes decreases by 23.3%. Therefore, the impregnation method can improve the quality of carbon nanotubes in the case of a similar degree of difficulty in the preparation of the catalyst.

Expansion of large-scale tree plantations for commodity crop and timber production is a leading cause of tropical deforestation. While automated detection of plantations across large spatial scales and with high temporal resolution is critical to inform policies to reduce deforestation, such mapping is technically challenging. Thus, most available plantation maps rely on visual inspection of imagery, and many of them are limited to small areas for specific years. Here, we present an automated approach, which we call Plantation Analysis by Learning from Multiple Classes (PALM), for mapping plantations on an annual basis using satellite remote sensing data. Due to the heterogeneity of land cover classes, PALM utilizes ensemble learning to simultaneously incorporate training samples from multiple land cover classes over different years. After the ensemble learning, we further improve the performance by post-processing using a Hidden Markov Model. We implement the proposed automated approach using MODIS data in Sumatra and Indonesian Borneo (Kalimantan). To validate the classification, we compare plantations detected using our approach with existing datasets developed through visual interpretation. Based on random sampling and comparison with high-resolution images, the user’s accuracy and producer’s accuracy of our generated map are around 85% and 80% in our study region.

Formaldehyde (FA) is a well-known irritant, and it is suggested to increase the risk of immune diseases and cancer. The present study aimed to evaluate the distribution of major lymphocyte subsets and cytokine expression profiles in the peripheral blood of FA-exposed workers. A total of 118 FA-exposed workers and 79 controls were enrolled in the study. High performance liquid chromatography, flow cytometry, and cytometric bead array were used to analyze FA in air sample and formic acid in urine, blood lymphocyte subpopulations, and serum cytokines, respectively. The FA-exposed workers were divided into low and high exposure groups according to their exposure levels. The results showed that both the low and high FA-exposed groups had a significant increase of formic acid in urine when compared to the controls. Both the low and high exposure groups had a significant increase in the percentage of B cells (CD19+) compared to the control group (p