Ensuring the safety and efficacy of chemical compounds is crucial in small-molecule drug development. In the later stages of drug development, toxic compounds pose a significant challenge, losing valuable resources and time. Early and accurate prediction of compound toxicity using deep learning models offers a promising solution to mitigate these risks during drug discovery. In this study, we present the development of several deep-learning models aimed at evaluating different types of compound toxicity, including acute toxicity, carcinogenicity, hERG_cardiotoxicity (the human ether-a-go-go related gene caused cardiotoxicity), hepatotoxicity, and mutagenicity. To address the inherent variations in data size, label type, and distribution across different types of toxicity, we employed diverse training strategies. Our first approach involved utilizing a graph convolutional network (GCN) regression model to predict acute toxicity, which achieved notable performance with Pearson R 0.76, 0.74, and 0.65 for intraperitoneal, intravenous, and oral administration routes, respectively. Furthermore, we trained multiple GCN binary classification models, each tailored to a specific type of toxicity. These models exhibited high area under the curve (AUC) scores, with an impressive AUC of 0.69, 0.77, 0.88, and 0.79 for predicting carcinogenicity, hERG_cardiotoxicity, mutagenicity, and hepatotoxicity, respectively. Additionally, we have used the approved drug dataset to determine the appropriate threshold value for the prediction score in model usage. We integrated these models into a virtual screening pipeline to assess their effectiveness in identifying potential low-toxicity drug candidates. Our findings indicate that this deep learning approach has the potential to significantly reduce the cost and risk associated with drug development by expediting the selection of compounds with low toxicity profiles. Therefore, the models developed in this study hold promise as critical tools for early drug candidate screening and selection.

The extensive use of plastics leads to the release and diffusion of microplastics. Household plastic products occupy a large part and are closely related to daily life. Due to the small size and complex composition of microplastics, it is challenging to identify and quantify microplastics. Therefore,a multi-model machine learning approach was developed for classification of household microplastics based on Raman spectroscopy. In this study, Raman spectroscopy and machine learning algorithm are combined to realize the accurate identification of seven standard microplastic samples, real microplastics samples and real microplastic samples post-exposure to environmental stresses. Four single-model machine learning methods were used in this study, including Support vector machine (SVM), K-nearest neighbor (KNN), Linear discriminant analysis (LDA), and Multi-layer perceptron (MLP) model. The principal components analysis (PCA) was utilized before SVM, KNN and LDA. The classification effect of four models on standard plastic samples is over 88%, and reliefF algorithm was used to distinguish HDPE and LDPE samples. A multi-model is proposed based on four single models including PCA-LDA, PCA-KNN and MLP. The recognition accuracy of multi-model for standard microplastic samples, real microplastic samples and microplastic samples post-exposure to environmental stresses is over 98%. Our study demonstrates that the multi-model coupled with Raman spectroscopy is a valuable tool for microplastic classification.

Big data and data analysis methods and models are important tools in food security (FS) studies for gap analysis and preparation of appropriate analytical frameworks. These innovations necessitate the development of novel methods for collecting, storing, processing, and extracting data.
The primary goal of this study was to conduct a critical review of agricultural big data and methods and models used for FS studies published in peer-reviewed journals since 2010. Approximately 130 articles were selected for full content review after the pre-screening process.
There are different sources of data collection, including but not limited to online databases, the internet, omics, Internet of Things, social media, survey rounds, remote sensing, and the Food and Agriculture Organization Corporate Statistical Database. The collected data require analysis (i.e., mining, neural networks, Bayesian networks, and other ML algorithms) before data visualization using Python, R, Circos, Gephi, Tableau, or Cytoscape. Approximately 122 models, all of which were used in FS studies worldwide, were selected from 130 articles. However, most of these models addressed only one or two dimensions of FS (i.e., availability and access) and ignored the other dimensions (i.e., stability and utilization), creating a gap in the global context.
There are certain FS gaps both worldwide and in the United Arab Emirates that need to be addressed by scientists and policymakers. Following the identification of the drivers, policies, and indicators, the findings of this review could be used to develop an appropriate analytical framework for FS and nutrition.

Insightful understanding of size-dependent optical signatures and precise regularity of nanosensors is critical for developing applications of plasmonic sensing. This work presents a systematic study on localized surface plasmon resonance (LSPR)-based nanosensors of plasmonic silver nanocubes (AgNCs) with the edge lengths of 59.84 ± 7.97 nm (no. 1 AgNCs), 75.70 ± 9.05 nm (no. 2 AgNCs) and 110.32 ± 14.63 nm (no. 3 AgNCs), respectively. The effects of different sizes on the scattering signatures and refractive index (RI) sensitivities of AgNCs were in situ determined using the multi-model co-localization approach of single AgNC by dark-field microscope (DFM), LSPR spectroscopy and scanning electron microscopy (SEM). The scattering light colour of single AgNC took place bathochromic shift from monocolour to multicolour with the growth of edge length of single AgNC. The LSPR scattering spectra of no. 1 and 2 AgNCs exhibited singlet and singlet with the shoulder peak from quadrupolar resonance mode, respectively. Compared with the scattering signatures of no. 1 and 2 AgNCs, the interesting LSPR effect of plasmon line shape with two distinct peaks was observed on single no. 3 AgNC. In situ studies on the scattering spectral response of single AgNC to the ambient solvents and probing the small-molecule adsorbates on the surface of single silver nanocube reveal that no. 2 AgNC is more suitable as nanosensor due to strong regularity and higher sensitivity. The mechanism involved in optical signatures was elaborated clearly by combining with the experiments and theoretical simulation.

The geological conceptual model is considered a major source of uncertainty in groundwater modelling and well capture zone delineation. However, how to account for it in groundwater policy and management remains largely unresolved. We explore the drivers and barriers to account for geological conceptual uncertainty in groundwater protection amongst decision makers and stakeholders in an agricultural groundwater catchment in Denmark. Using a groundwater model, we analyze the impact of alternative geological conceptual models on capture zone delineation. A focus area, which covers multiple modelled capture zones, is defined and considered for groundwater protection. Model uncertainty and focus area are discussed at two workshops, one with local and national stakeholders and another with local farmers. The drivers to account for model uncertainty include: i) safer drinking water protection by considering a larger area for protection than identified from a single geological model; and ii) stability over time of management plans. The main barrier is the additional cost to the stakeholders for the protection of a larger area. We conclude that integration of geological uncertainty in groundwater protection plans may be improved through: i) better communication between the research community and the national water authority; ii) more constraining guidelines regarding the estimation of geological uncertainty; and iii) the development of a framework ensuring knowledge transfer to the local water authorities and detailing how to integrate uncertainty in management plans.

Although the world understands the possible threat of the future of climate changes, there remain serious barriers to be resolved in terms of policy decisions. The scientific and the societal uncertainties in the climate change policies must be the large part of this barrier. Following the Paris Agreement, the world comes to the next stage to decide the next actions. Without a view of risk management, any decision will be \"based on neglecting alternatives\" behavior. The Ministry of the Environment, Japan has established an inter-disciplinary research project, called Integrated Climate Assessment-Risks, Uncertainties, and Society (ICA-RUS) conducted by Dr. Seita Emori, National Institute for Environmental Studies. ICA-RUS consists of five research themes, i.e., (1) synthesis of global climate risks, (2) optimization of land, water, and ecosystem for climate risks, (3) analysis of critical climate risks, (4) evaluation of climate risk management options, and (5) interactions between scientific and social rationalities. We participated in the fourth theme to provide the quantitative assessment of technology options and policy measures by integrating assessment model simulations. We employ the multi-model approach to deal with the complex relationships among various fields such as technology, economics, and land use changes. Four different types of integrated assessment models, i.e., MARIA-14 (Mori), EMEDA (Washida), GRAPE (Kurosawa), and AIM (Masui), participate in the fourth research theme. These models contribute to the ICA-RUS by providing two information categories. First, these models provide common simulation results based on shared socioeconomic pathway scenarios and the shared climate policy cases given by the first theme of ICA-RUS to see the ranges of the evaluation. Second, each model also provides model-specific outcomes to answer special topics, e.g., geoengineering, sectoral trade, adaptation, and decision making under uncertainties. The purpose of this paper is to describe the outline and the main outcomes of the multi-model inter-comparison among the four models with a focus upon the first and to present the main outcomes. Furthermore, in this study, we introduce a statistical meta-analysis of the multi-model simulation results to see whether the differently structured models provide the inter-consistent findings. The major findings of our activities are as follows: First, in the stringent climate target, the regional economic losses among models tend to diverge, whereas global total economic loss does not. Second, both carbon capture and storage (CCS) as well as BECCS are essential for providing the feasibility of stringent climate targets even if the deployment potential varies among models. Third, the models show small changes in the crop production in world total, whereas large differences appear between regions. Fourth, the statistical meta-analysis of the multi-model simulation results suggests that the models would have an implicit but common relationship between gross domestic product losses and mitigation options even if their structures and simulation results are different. Since this study is no more than a preliminary exercise of the statistical meta-analysis, it is expected that more sophisticated methods such as data mining or machine learning could be applicable to the simulation database to extract the implicit information behind the models.

Overcome the lack of enough samples in gene expression data sets having thousands of genes but a small number of samples challenging the computational methods using them.
This paper introduces a multi-model artificial gene expression data generation framework where different gene regulatory network (GRN) models contribute to the final set of samples based on the characteristics of their underlying paradigms. In the first stage, we build different GRN models, and sample data from each of them separately. Then, we pool the generated samples into a rich set of gene expression samples, and finally try to select the best of the generated samples based on a multi-objective selection method measuring the quality of the generated samples from three different aspects such as compatibility, diversity and coverage. We use four alternative GRN models, namely, ordinary differential equations, probabilistic Boolean networks, multi-objective genetic algorithm and hierarchical Markov model.
We conducted a comprehensive set of experiments based on both real-life biological and synthetic gene expression data sets. We show that our multi-objective sample selection mechanism effectively combines samples from different models having up to 95% compatibility, 10% diversity and 50% coverage. We show that the samples generated by our framework has up to 1.5x higher compatibility, 2x higher diversity and 2x higher coverage than the samples generated by the individual models that the multi-model framework uses. Moreover, the results show that the GRNs inferred from the samples generated by our framework can have 2.4x higher precision, 12x higher recall, and 5.4x higher f-measure values than the GRNs inferred from the original gene expression samples.
Therefore, we show that, we can significantly improve the quality of generated gene expression samples by integrating different computational models into one unified framework without dealing with complex internal details of each individual model. Moreover, the rich set of artificial gene expression samples is able to capture some biological relations that can even not be captured by the original gene expression data set.

This paper deals with characterization and modelling of human handwriting motion from two forearm muscle activity signals, called electromyography signals (EMG). In this work, an experimental approach was used to record the coordinates of a pen tip moving on the (x, y) plane and EMG signals during the handwriting act. The main purpose is to design a new mathematical model which characterizes this biological process. Based on a multi-model approach, this system was originally developed to generate letters and geometric forms written by different writers. A Recursive Least Squares algorithm is used to estimate the parameters of each sub-model of the multi-model basis. Simulations show good agreement between predicted results and the recorded data.