[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Sign in to use this feature.

Years

Between: -

Subjects

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Journals

Article Types

Countries / Regions

Search Results (32)

Search Parameters:
Keywords = ELM database

Order results
Result details
Results per page
Select all
Export citation of selected articles as:
17 pages, 2446 KiB  
Article
The Impact of the SMOTE Method on Machine Learning and Ensemble Learning Performance Results in Addressing Class Imbalance in Data Used for Predicting Total Testosterone Deficiency in Type 2 Diabetes Patients
by Mehmet Kivrak, Ugur Avci, Hakki Uzun and Cuneyt Ardic
Diagnostics 2024, 14(23), 2634; https://doi.org/10.3390/diagnostics14232634 - 22 Nov 2024
Viewed by 523
Abstract
Background and Objective: Diabetes Mellitus is a long-term, multifaceted metabolic condition that necessitates ongoing medical management. Hypogonadism is a syndrome that is a clinical and/or biochemical indicator of testosterone deficiency. Cross-sectional studies have reported that 20–80.4% of all men with Type 2 diabetes [...] Read more.
Background and Objective: Diabetes Mellitus is a long-term, multifaceted metabolic condition that necessitates ongoing medical management. Hypogonadism is a syndrome that is a clinical and/or biochemical indicator of testosterone deficiency. Cross-sectional studies have reported that 20–80.4% of all men with Type 2 diabetes have hypogonadism, and Type 2 diabetes is related to low testosterone. This study presents an analysis of the use of ML and EL classifiers in predicting testosterone deficiency. In our study, we compared optimized traditional ML classifiers and three EL classifiers using grid search and stratified k-fold cross-validation. We used the SMOTE method for the class imbalance problem. Methods: This database contains 3397 patients for the assessment of testosterone deficiency. Among these patients, 1886 patients with Type 2 diabetes were included in the study. In the data preprocessing stage, firstly, outlier/excessive observation analyses were performed with LOF and missing value analyses were performed with random forest. The SMOTE is a method for generating synthetic samples of the minority class. Four basic classifiers, namely MLP, RF, ELM and LR, were used as first-level classifiers. Tree ensemble classifiers, namely ADA, XGBoost and SGB, were used as second-level classifiers. Results: After the SMOTE, while the diagnostic accuracy decreased in all base classifiers except ELM, sensitivity values increased in all classifiers. Similarly, while the specificity values decreased in all classifiers, F1 score increased. The RF classifier gave more successful results on the base-training dataset. The most successful ensemble classifier in the training dataset was the ADA classifier in the original data and in the SMOTE data. In terms of the testing data, XGBoost is the most suitable model for your intended use in evaluating model performance. XGBoost, which exhibits a balanced performance especially when the SMOTE is used, can be preferred to correct class imbalance. Conclusions: The SMOTE is used to correct the class imbalance in the original data. However, as seen in this study, when the SMOTE was applied, the diagnostic accuracy decreased in some models but the sensitivity increased significantly. This shows the positive effects of the SMOTE in terms of better predicting the minority class. Full article
(This article belongs to the Section Machine Learning and Artificial Intelligence in Diagnostics)
Show Figures

Figure 1

Figure 1
<p>Testesterone target organs [<a href="#B7-diagnostics-14-02634" class="html-bibr">7</a>].</p>
Full article ">Figure 2
<p>The working step.</p>
Full article ">Figure 3
<p>Outlier/excessive observation analyses with local outlier factor. The observations shown in blue in the figure are values within normal limits. The values above the red line are outliers.</p>
Full article ">Figure 4
<p>Illustration of class imbalance.</p>
Full article ">Figure 5
<p>Original and preprocessed (SMOTE) data. Blue color normal individuals and green color testosterone deficiency (TD) individuals.</p>
Full article ">Figure 6
<p>Classification diagram for original and SMOTE data using base classifiers (training data). (<b>A1</b>): Original data of MLP, (<b>A2</b>): SMOTE data of MLP, (<b>B1</b>): Original data of RF, (<b>B2</b>): SMOTE data of RF, (<b>C1</b>): Original data of LR, (<b>C2</b>): SMOTE data of LR, (<b>D1</b>): Original data of ELM, (<b>D2</b>): SMOTE data of ELM.</p>
Full article ">Figure 7
<p>Classification diagram for original and SMOTE data using base classifiers (training data). (<b>A1</b>): Original data of ADA, (<b>A2</b>): SMOTE data of ADA, (<b>B1</b>): Original data of XGBoost, (<b>B2</b>): SMOTE data of XGBoost, (<b>C1</b>): Original data of SGB, (<b>C2</b>): SMOTE data of SGB.</p>
Full article ">
14 pages, 3633 KiB  
Article
Delineating Regional BES–ELM Neural Networks for Studying Indoor Visible Light Positioning
by Jiaming Zhang and Xizheng Ke
Photonics 2024, 11(10), 910; https://doi.org/10.3390/photonics11100910 - 27 Sep 2024
Viewed by 644
Abstract
This paper introduces a single LED and four photodetectors (PDs) as a visible light system structure and collects the received signal strength values and corresponding physical coordinates at the PD receiving end, establishing a comprehensive dataset. The K-means clustering algorithm is employed to [...] Read more.
This paper introduces a single LED and four photodetectors (PDs) as a visible light system structure and collects the received signal strength values and corresponding physical coordinates at the PD receiving end, establishing a comprehensive dataset. The K-means clustering algorithm is employed to separate the room into center and boundary areas through the fingerprint database. The bald eagle search (BES) algorithm is employed to optimize the initial parameters, specifically the weights and thresholds, in the extreme learning machine (ELM) neural network, and the BES–ELM indoor positioning model is established by region to improve positioning accuracy. Due to the impact exerted by the ambient environment, there are fluctuations in the positioning accuracy of the center and edge regions, and the positioning of the edge region needs to be further improved. To address this, it is proposed to use the enhanced weighted K-nearest neighbor (EWKNN) algorithm based on the BES–ELM neural network to correct the prediction points with higher-than-average positioning errors, achieving precise edge positioning. The simulation demonstrates that within an indoor space measuring 5 m × 5 m × 3 m, the algorithm achieves an average positioning error of 2.93 cm, and the positioning accuracy is improved by 86.07% relative to conventional BP neural networks. Full article
Show Figures

Figure 1

Figure 1
<p>Indoor VLC system model.</p>
Full article ">Figure 2
<p>Indoor visible light channel [<a href="#B21-photonics-11-00910" class="html-bibr">21</a>].</p>
Full article ">Figure 3
<p>ELM neural network structure.</p>
Full article ">Figure 4
<p>The flow of the BES–ELM positioning algorithm.</p>
Full article ">Figure 5
<p>Dispersion of the optical power received. (<b>a</b>) LOS channel; (<b>b</b>) NLOS channel; (<b>c</b>) total channel.</p>
Full article ">Figure 6
<p>Regionalization results.</p>
Full article ">Figure 7
<p>Comparison plots before and after correction of the edge region. (<b>a</b>) Comparison map before correction; (<b>b</b>) Comparison map after screening; (<b>c</b>) Comparison map after correction; (<b>d</b>) Global comparison map of the edge region after being updated.</p>
Full article ">Figure 8
<p>Average positioning error for different numbers of neurons.</p>
Full article ">Figure 9
<p>Positioning error distribution: (<b>a</b>) Positioning error distribution of ELM; (<b>b</b>) Positioning error distribution of BES–ELM; (<b>c</b>) Positioning error distribution of edge-corrected BES–ELM.</p>
Full article ">Figure 10
<p>Comparison plots of predicted and true coordinates: (<b>a</b>) ELM comparison plot; (<b>b</b>) BES–ELM comparison plot; (<b>c</b>) BES–ELM comparison plot after edge correction.</p>
Full article ">Figure 11
<p>Distribution of positioning errors throughout time.</p>
Full article ">
24 pages, 4234 KiB  
Article
Breast Cancer Detection and Analytics Using Hybrid CNN and Extreme Learning Machine
by Vidhushavarshini Sureshkumar, Rubesh Sharma Navani Prasad, Sathiyabhama Balasubramaniam, Dhayanithi Jagannathan, Jayanthi Daniel and Seshathiri Dhanasekaran
J. Pers. Med. 2024, 14(8), 792; https://doi.org/10.3390/jpm14080792 - 26 Jul 2024
Cited by 2 | Viewed by 2318
Abstract
Early detection of breast cancer is essential for increasing survival rates, as it is one of the primary causes of death for women globally. Mammograms are extensively used by physicians for diagnosis, but selecting appropriate algorithms for image enhancement, segmentation, feature extraction, and [...] Read more.
Early detection of breast cancer is essential for increasing survival rates, as it is one of the primary causes of death for women globally. Mammograms are extensively used by physicians for diagnosis, but selecting appropriate algorithms for image enhancement, segmentation, feature extraction, and classification remains a significant research challenge. This paper presents a computer-aided diagnosis (CAD)-based hybrid model combining convolutional neural networks (CNN) with a pruned ensembled extreme learning machine (HCPELM) to enhance breast cancer detection, segmentation, feature extraction, and classification. The model employs the rectified linear unit (ReLU) activation function to enhance data analytics after removing artifacts and pectoral muscles, and the HCPELM hybridized with the CNN model improves feature extraction. The hybrid elements are convolutional and fully connected layers. Convolutional layers extract spatial features like edges, textures, and more complex features in deeper layers. The fully connected layers take these features and combine them in a non-linear manner to perform the final classification. ELM performs classification and recognition tasks, aiming for state-of-the-art performance. This hybrid classifier is used for transfer learning by freezing certain layers and modifying the architecture to reduce parameters, easing cancer detection. The HCPELM classifier was trained using the MIAS database and evaluated against benchmark methods. It achieved a breast image recognition accuracy of 86%, outperforming benchmark deep learning models. HCPELM is demonstrating superior performance in early detection and diagnosis, thus aiding healthcare practitioners in breast cancer diagnosis. Full article
(This article belongs to the Section Methodology, Drug and Device Discovery)
Show Figures

Figure 1

Figure 1
<p>Raw mammogram image and pectoral muscle removed image.</p>
Full article ">Figure 2
<p>Before and after removal of pectoral muscles.</p>
Full article ">Figure 3
<p>Pectoral muscle removal.</p>
Full article ">Figure 4
<p>Results of pectoral muscle removal.</p>
Full article ">Figure 5
<p>Accuracy of individual versus ensemble ELM.</p>
Full article ">Figure 6
<p>Training versus validation accuracy of the proposed HCPELM model.</p>
Full article ">Figure 7
<p>Confusion matrix for HCPELM model.</p>
Full article ">Figure 8
<p>ROC for HCPELM model.</p>
Full article ">Figure 9
<p>Comparison of proposed HCPELM and other models.</p>
Full article ">Figure 10
<p>McNemar test results.</p>
Full article ">
18 pages, 5839 KiB  
Article
Enhancing Road Safety: Deep Learning-Based Intelligent Driver Drowsiness Detection for Advanced Driver-Assistance Systems
by Eunmok Yang and Okyeon Yi
Electronics 2024, 13(4), 708; https://doi.org/10.3390/electronics13040708 - 9 Feb 2024
Cited by 4 | Viewed by 4365
Abstract
Driver drowsiness detection is a significant element of Advanced Driver-Assistance Systems (ADASs), which utilize deep learning (DL) methods to improve road safety. A driver drowsiness detection system can trigger timely alerts like auditory or visual warnings, thereby stimulating drivers to take corrective measures [...] Read more.
Driver drowsiness detection is a significant element of Advanced Driver-Assistance Systems (ADASs), which utilize deep learning (DL) methods to improve road safety. A driver drowsiness detection system can trigger timely alerts like auditory or visual warnings, thereby stimulating drivers to take corrective measures and ultimately avoiding possible accidents caused by impaired driving. This study presents a Deep Learning-based Intelligent Driver Drowsiness Detection for Advanced Driver-Assistance Systems (DLID3-ADAS) technique. The DLID3-ADAS technique aims to enhance road safety via the detection of drowsiness among drivers. Using the DLID3-ADAS technique, complex features from images are derived through the use of the ShuffleNet approach. Moreover, the Northern Goshawk Optimization (NGO) algorithm is exploited for the selection of optimum hyperparameters for the ShuffleNet model. Lastly, an extreme learning machine (ELM) model is used to properly detect and classify the drowsiness states of drivers. The extensive set of experiments conducted based on the Yawdd driver database showed that the DLID3-ADAS technique achieves a higher performance compared to existing models, with a maximum accuracy of 97.05% and minimum computational time of 0.60 s. Full article
(This article belongs to the Special Issue Signal Processing and AI Applications for Vehicles)
Show Figures

Figure 1

Figure 1
<p>The overall procedure of the DLID3-ADAS system.</p>
Full article ">Figure 2
<p>ELM structure.</p>
Full article ">Figure 3
<p>(<b>a</b>) Drowsiness and (<b>b</b>) non-drowsiness images.</p>
Full article ">Figure 4
<p>(<b>a</b>,<b>b</b>) Confusion matrices based on a ratio of 80:20 for TRPH/TSPH; (<b>c</b>) PR curve based on a ratio of 80:20 for TRPH/TSPH; and (<b>d</b>) ROC curve based on a ratio of 80:20 for TRPH/TSPH.</p>
Full article ">Figure 5
<p>The average outcome of the DLID3-ADAS system under a ratio of 80:20 for TRPH/TSPH.</p>
Full article ">Figure 6
<p><math display="inline"><semantics> <mrow> <mi>A</mi> <mi>c</mi> <mi>c</mi> <msub> <mi>u</mi> <mi>y</mi> </msub> <mo> </mo> </mrow> </semantics></math> curves of the DLID3-ADAS model under a ratio of 80:20 for TRPH/TSPH.</p>
Full article ">Figure 7
<p>Loss curve of the DLID3-ADAS technique under a ratio of 80:20 for TRPH/TSPH.</p>
Full article ">Figure 8
<p>(<b>a</b>,<b>b</b>) Confusion matrices with a ratio of 70:30 for TRPH/TSPH; (<b>c</b>) PR curve under a ratio of 70:30 for TRPH/TSPH; and (<b>d</b>) ROC curve under a ratio of 70:30 for TRPH/TSPH.</p>
Full article ">Figure 9
<p>The average outcome of the DLID3-ADAS algorithm under a ratio of 70:30 for TRPH/TSPH.</p>
Full article ">Figure 10
<p><math display="inline"><semantics> <mrow> <mi>A</mi> <mi>c</mi> <mi>c</mi> <msub> <mi>u</mi> <mi>y</mi> </msub> </mrow> </semantics></math> curves of the DLID3-ADAS model under a ratio of 70:30 for TRPH/TSPH.</p>
Full article ">Figure 11
<p>Loss curves of the DLID3-ADAS method under a ratio of 70:30 for TRPH/TSPH.</p>
Full article ">Figure 12
<p>Comparison analysis of the DLID3-ADAS technique with other models.</p>
Full article ">Figure 13
<p>CT analysis of the DLID3-ADAS technique compared to other models.</p>
Full article ">
14 pages, 2779 KiB  
Article
Predicting the Temperature-Dependent Long-Term Creep Mechanical Response of Silica Sand-Textured Geomembrane Interfaces Based on Physical Tests and Machine Learning Techniques
by Zhiming Chao, Haoyu Wang, Hanwen Hu, Tianchen Ding and Ye Zhang
Materials 2023, 16(18), 6144; https://doi.org/10.3390/ma16186144 - 10 Sep 2023
Cited by 3 | Viewed by 1600
Abstract
Preciously assessing the creep mechanical response of sand–geomembrane interfaces is vital for the design of relevant engineering applications, which is inevitable to be influenced by temperature and stress statuses. In this paper, based on the self-developed temperature-controlled large interface shear apparatus, a series [...] Read more.
Preciously assessing the creep mechanical response of sand–geomembrane interfaces is vital for the design of relevant engineering applications, which is inevitable to be influenced by temperature and stress statuses. In this paper, based on the self-developed temperature-controlled large interface shear apparatus, a series of long-term creep shear tests on textured geomembrane–silica sand interfaces in different temperatures, normal pressure, and creep shear pressure were conducted, and a database compiled from the physical creep shear test results is constructed. By adopting the database, three disparate machine learning algorithms of the Back Propagation Artificial Neural Network (BPANN), the Support Vector Machine (SVM) and the Extreme Learning Machine (ELM) were adopted to assess the long-term creep mechanical properties of sand–geomembrane interfaces while also considering the influence of temperature. Then, the forecasting results of the different algorithms was compared and analyzed. Furthermore, by using the optimal machine learning model, sensitivity analysis was carried out. The research indicated that the BPANN model has the best forecasting performance according to the statistics criteria of the Root-Mean-Square Error, the Correlation Coefficient, Wilmot’s Index of Agreement, and the Mean Absolute Percentage Error among the developed models. Temperature is the most important influence factor on the creep interface mechanical properties, followed with time. The research findings can support the operating safety of the related engineering facilities installed with the geomembrane. Full article
Show Figures

Figure 1

Figure 1
<p>The tested textured geomembrane.</p>
Full article ">Figure 2
<p>The schematic diagram of BPANN model.</p>
Full article ">Figure 3
<p>Data distribution of the complied database. (<b>a</b>) Normal pressure; (<b>b</b>) creep shear pressure; (<b>c</b>) temperature; and (<b>d</b>) time.</p>
Full article ">Figure 4
<p>The optimization process. (<b>a</b>) BPANN; and (<b>b</b>) ELM.</p>
Full article ">Figure 5
<p>RMSE value.</p>
Full article ">Figure 6
<p>MAPE value.</p>
Full article ">Figure 7
<p>WI value.</p>
Full article ">Figure 8
<p>R<sup>2</sup> value.</p>
Full article ">Figure 9
<p>The sensitivity analysis results.</p>
Full article ">Figure 10
<p>The measured and predicted creep shear displacement.</p>
Full article ">
23 pages, 5380 KiB  
Article
SoftVein-WELM: A Weighted Extreme Learning Machine Model for Soft Biometrics on Palm Vein Images
by David Zabala-Blanco, Ruber Hernández-García and Ricardo J. Barrientos
Electronics 2023, 12(17), 3608; https://doi.org/10.3390/electronics12173608 - 26 Aug 2023
Cited by 4 | Viewed by 1410
Abstract
Contactless biometric technologies such as palm vein recognition have gained more relevance in the present and immediate future due to the COVID-19 pandemic. Since certain soft biometrics like gender and age can generate variations in the visualization of palm vein patterns, these soft [...] Read more.
Contactless biometric technologies such as palm vein recognition have gained more relevance in the present and immediate future due to the COVID-19 pandemic. Since certain soft biometrics like gender and age can generate variations in the visualization of palm vein patterns, these soft traits can reduce the penetration rate on large-scale databases for mass individual recognition. Due to the limited availability of public databases, few works report on the existing approaches to gender and age classification through vein pattern images. Moreover, soft biometric classification commonly faces the problem of imbalanced data class distributions, representing a limitation of the reported approaches. This paper introduces weighted extreme learning machine (W-ELM) models for gender and age classification based on palm vein images to address imbalanced data problems, improving the classification performance. The highlights of our proposal are that it avoids using a feature extraction process and can incorporate a weight matrix in optimizing the ELM model by exploiting the imbalanced nature of the data, which guarantees its application in realistic scenarios. In addition, we evaluate a new class distribution for soft biometrics on the VERA dataset and a new multi-label scheme identifying gender and age simultaneously. The experimental results demonstrate that both evaluated W-ELM models outperform previous existing approaches and a novel CNN-based method in terms of the accuracy and G-mean metrics, achieving accuracies of 98.91% and 99.53% for gender classification on VERA and PolyU, respectively. In more challenging scenarios for age and gender–age classifications on the VERA dataset, the proposed method reaches accuracies of 97.05% and 96.91%, respectively. The multi-label classification results suggest that further studies can be conducted on multi-task ELM for palm vein recognition. Full article
Show Figures

Figure 1

Figure 1
<p>Overview of the proposed SoftVein-WELM model for gender and age single/multi-label classification on palm vein images using weighted ELM.</p>
Full article ">Figure 2
<p>Overall distribution of age and sex groups from the VERA database.</p>
Full article ">Figure 3
<p>Palmprint images samples from the NIR spectrum of PolyU dataset.</p>
Full article ">Figure 4
<p>Sample augmentation process adopted from [<a href="#B26-electronics-12-03608" class="html-bibr">26</a>]. A central sliding window is translated by 5 px on both axes until the ROI border and also rotated by 5° with respect to the center of the palm until <math display="inline"><semantics> <mrow> <mo>±</mo> <mn>15</mn> <mo>°</mo> </mrow> </semantics></math>.</p>
Full article ">Figure 5
<p>Performance in terms of the number of hidden neurons and regularization parameter of the (<b>a</b>) W1-ELM and (<b>b</b>) W2-ELM models for gender classification task (two classes) on the VERA dataset.</p>
Full article ">Figure 6
<p>Performance in terms of the number of hidden neurons and regularization parameter by the (<b>a</b>) W1-ELM and (<b>b</b>) W2-ELM models for gender classification task (two classes) on the PolyU dataset.</p>
Full article ">Figure 7
<p>Performance in terms of the number of hidden neurons and regularization parameter of the (<b>a</b>) W1-ELM and (<b>b</b>) W2-ELM models for age classification task (four classes) on the VERA dataset.</p>
Full article ">Figure 8
<p>Performance in terms of the number of hidden neurons and regularization parameter of the (<b>a</b>) W1-ELM and (<b>b</b>) W2-ELM models for gender–age classification task (eight classes) on the VERA dataset.</p>
Full article ">Figure 9
<p>Examples of the confusion matrices for W2-ELM on both multi-class tasks: (<b>a</b>) age and (<b>b</b>) gender–age classification. Each row of the matrices corresponds to the true class, and each column indicates the predicted class. For gender–age classification, M represents “male” and F represents “female” in the label names. Maps of different colors are used for low and high values to better distinguish between the closer values.</p>
Full article ">
27 pages, 5814 KiB  
Article
Global Water Quality of Inland Waters with Harmonized Landsat-8 and Sentinel-2 Using Cloud-Computed Machine Learning
by Leonardo F. Arias-Rodriguez, Ulaş Firat Tüzün, Zheng Duan, Jingshui Huang, Ye Tuo and Markus Disse
Remote Sens. 2023, 15(5), 1390; https://doi.org/10.3390/rs15051390 - 1 Mar 2023
Cited by 22 | Viewed by 6038
Abstract
Modeling inland water quality by remote sensing has already demonstrated its capacity to make accurate predictions. However, limitations still exist for applicability in diverse regions, as well as to retrieve non-optically active parameters (nOAC). Models are usually trained only with water samples from [...] Read more.
Modeling inland water quality by remote sensing has already demonstrated its capacity to make accurate predictions. However, limitations still exist for applicability in diverse regions, as well as to retrieve non-optically active parameters (nOAC). Models are usually trained only with water samples from individual or local groups of waterbodies, which limits their capacity and accuracy in predicting parameters across diverse regions. This study aims to increase data availability to understand the performance of models trained with heterogeneous databases from both remote sensing and field measurement sources to improve machine learning training. This paper seeks to build a dataset with worldwide lake characteristics using data from water monitoring programs around the world paired with harmonized data of Landsat-8 and Sentinel-2. Additional feature engineering is also examined. The dataset is then used for model training and prediction of water quality at the global scale, time series analysis and water quality maps for lakes in different continents. Additionally, the modeling performance of nOACs are also investigated. The results show that trained models achieve moderately high correlations for SDD, TURB and BOD (R2 = 0.68) but lower performances for TSM and NO3-N (R2 = 0.43). The extreme learning machine (ELM) and the random forest regression (RFR) demonstrate better performance. The results indicate that ML algorithms can process remote sensing data and additional features to model water quality at the global scale and contribute to address the limitations of transferring and retrieving nOAC. However, significant limitations need to be considered, such as calibrated harmonization of water data and atmospheric correction procedures. Moreover, further understanding of the mechanisms that facilitate nOAC prediction is necessary. We highlight the need for international contributions to global water quality datasets capable of providing extensive water data for the improvement of global water monitoring. Full article
Show Figures

Figure 1

Figure 1
<p>The global location of all the stations from the above-mentioned data sources in raw form.</p>
Full article ">Figure 2
<p>Overview of the HLS processing.</p>
Full article ">Figure 3
<p>Sum and individual correlations of the water quality parameters with predicting features. The total of each node represents the sum of absolute value of positive and negative correlations between all the parameters with all predictors.</p>
Full article ">Figure 4
<p>Comprehensive evaluation of tested algorithms based on the relevant error metrics for optimal performance. The algorithms use the best source dataset in all cases.</p>
Full article ">Figure 5
<p>Scatterplots of modeled and measured water quality parameters in the test dataset.</p>
Full article ">Figure 6
<p>Train and test average <span class="html-italic">R</span><sup>2</sup> for each algorithm and dataset. In (<b>a</b>,<b>b</b>) improvement is noticeable when using datasets that have the RT features which are colored in red for both train and test phases. Similarly, the increase in the performance is seen on all the models when using an HBRT or FERT dataset, (<b>c</b>,<b>d</b>).</p>
Full article ">Figure 7
<p>Time series and spatial distribution of (<b>a</b>) Chl-a in Lake Tahoe (U.S., 29 November 2021), (<b>b</b>) DO in Lake Vichuquen (Chile, 29 November 2021) and (<b>c</b>) SDD for Lake Trasimeno (Italy, 31 August 2021). Background image: harmonized red band in greyscale. The plots show the average of the parameter for the whole lake. Spatial variation is visible in the maps.</p>
Full article ">Figure 8
<p>Sum and individual correlations of the OAC with nOAC. Diagrams of (<b>a</b>), (<b>b</b>) and (<b>c</b>) TSM, BOD and SDD, respectively.</p>
Full article ">Figure 9
<p>Average <span class="html-italic">R</span><sup>2</sup> for all the models by the nature of the target parameters. OAC: Chl-a, TURB, TSM and SDD. nOAC: DO, PTOT, NO3-N, BOD and COD.</p>
Full article ">Figure 10
<p>Individual correlation of each predictor with water quality parameters. Features are ranged from −1 to 1 depending on their higher positive or negative correlation. (<b>a</b>) displays correlations of predictors and targets. (<b>b</b>) fades the areas of very low or zero correlation (−0.20 ≤ <span class="html-italic">r</span> ≤ 0.20).</p>
Full article ">Figure 11
<p>Improvement of temporally and spatially aware models.</p>
Full article ">
19 pages, 1403 KiB  
Article
A Novel Hierarchical Extreme Machine-Learning-Based Approach for Linear Attenuation Coefficient Forecasting
by Giuseppe Varone, Cosimo Ieracitano, Aybike Özyüksel Çiftçioğlu, Tassadaq Hussain, Mandar Gogate, Kia Dashtipour, Bassam Naji Al-Tamimi, Hani Almoamari, Iskender Akkurt and Amir Hussain
Entropy 2023, 25(2), 253; https://doi.org/10.3390/e25020253 - 30 Jan 2023
Cited by 5 | Viewed by 2383
Abstract
The development of reinforced polymer composite materials has had a significant influence on the challenging problem of shielding against high-energy photons, particularly X-rays and γ-rays in industrial and healthcare facilities. Heavy materials’ shielding characteristics hold a lot of potential for bolstering concrete [...] Read more.
The development of reinforced polymer composite materials has had a significant influence on the challenging problem of shielding against high-energy photons, particularly X-rays and γ-rays in industrial and healthcare facilities. Heavy materials’ shielding characteristics hold a lot of potential for bolstering concrete chunks. The mass attenuation coefficient is the main physical factor that is utilized to measure the narrow beam γ-ray attenuation of various combinations of magnetite and mineral powders with concrete. Data-driven machine learning approaches can be investigated to assess the gamma-ray shielding behavior of composites as an alternative to theoretical calculations, which are often time- and resource-intensive during workbench testing. We developed a dataset using magnetite and seventeen mineral powder combinations at different densities and water/cement ratios, exposed to photon energy ranging from 1 to 1006 kiloelectronvolt (KeV). The National Institute of Standards and Technology (NIST) photon cross-section database and software methodology (XCOM) was used to compute the concrete’s γ-ray shielding characteristics (LAC). The XCOM-calculated LACs and seventeen mineral powders were exploited using a range of machine learning (ML) regressors. The goal was to investigate whether the available dataset and XCOM-simulated LAC can be replicated using ML techniques in a data-driven approach. The minimum absolute error (MAE), root mean square error (RMSE), and R2score were employed to assess the performance of our proposed ML models, specifically a support vector machine (SVM), 1d-convolutional neural network (CNN), multi-Layer perceptrons (MLP), linear regressor, decision tree, hierarchical extreme machine learning (HELM), extreme learning machine (ELM), and random forest networks. Comparative results showed that our proposed HELM architecture outperformed state-of-the-art SVM, decision tree, polynomial regressor, random forest, MLP, CNN, and conventional ELM models. Stepwise regression and correlation analysis were further used to evaluate the forecasting capability of ML techniques compared to the benchmark XCOM approach. According to the statistical analysis, the HELM model showed strong consistency between XCOM and predicted LAC values. Additionally, the HELM model performed better in terms of accuracy than the other models used in this study, yielding the highest R2score and the lowest MAE and RMSE. Full article
Show Figures

Figure 1

Figure 1
<p>Overview of the study’s methodology. In order to create the dataset, samples of mineral powders, magnetite, gamma energies, and densities were first chosen. Then, Panel (<b>a</b>) shows the preprocessing stage, which included out-layer removal and min-max normalization. The preprocessed data were then split into train (80%) and test (20%) groups at random. In Panel (<b>b</b>), the features vector is emphasized. Using the train dataset and the 10 k-fold cross-validation procedure, we trained our ML regressor models. To quantify the performance of the proposed ML regressors in Panel (<b>c</b>), we calculated regression and statistical analysis comparing the ML results to the XCOM simulated LAC.</p>
Full article ">Figure 2
<p>Flowchart of the proposed HELM architecture.</p>
Full article ">Figure 3
<p>Regression plots and correlation analysis on LAC data predicted by the proposed ML models.</p>
Full article ">Figure 4
<p>The figure presents the linear difference between input XCOM LAC and the LAC forecasted by the developed ML regressor models. Here x-axis represents the number of samples in the experiment, and y-axis represents the differences. Each panel also reports the mean and standard deviation.</p>
Full article ">
24 pages, 5681 KiB  
Article
E-Learning Performance Evaluation in Medical Education—A Bibliometric and Visualization Analysis
by Deborah Oluwadele, Yashik Singh and Timothy T. Adeliyi
Healthcare 2023, 11(2), 232; https://doi.org/10.3390/healthcare11020232 - 12 Jan 2023
Cited by 7 | Viewed by 2939
Abstract
Performance evaluation is one of the most critical components in assuring the comprehensive development of e-learning in medical education (e-LMED). Although several studies evaluate performance in e-LMED, no study presently maps the rising scientific knowledge and evolutionary patterns that establish a solid background [...] Read more.
Performance evaluation is one of the most critical components in assuring the comprehensive development of e-learning in medical education (e-LMED). Although several studies evaluate performance in e-LMED, no study presently maps the rising scientific knowledge and evolutionary patterns that establish a solid background to investigate and quantify the efficacy of the evaluation of performance in e-LMED. Therefore, this study aims to quantify scientific productivity, identify the key terms and analyze the extent of research collaboration in this domain. We searched the SCOPUS database using search terms informed by the PICOS model, and a total of 315 studies published between 1991 and 2022 were retrieved. Performance analysis, science mapping, network analysis, and visualization were performed using R Bibliometrix, Biblioshiny, and VOSviewer packages. Findings reveal that authors are actively publishing and collaborating in this domain, which experienced a sporadic publication increase in 2021. Most of the top publications, collaborations, countries, institutions, and journals are produced in first-world countries. In addition, studies evaluating performance in e-LMED evaluated constructs such as efficacy, knowledge gain, student perception, confidence level, acceptability, feasibility, usability, and willingness to recommend e-learning, mainly using pre-tests and post-tests experimental design methods. This study can help researchers understand the existing landscape of performance evaluation in e-LMED and could be used as a background to investigate and quantify the efficacy of the evaluation of e-LMED. Full article
Show Figures

Figure 1

Figure 1
<p>Documents by Type.</p>
Full article ">Figure 2
<p>The document by Discipline.</p>
Full article ">Figure 3
<p>Annual publication growth of Performance evaluation of e-LMED.</p>
Full article ">Figure 4
<p>Top publishing authors in Performance evaluation of e-LMED.</p>
Full article ">Figure 5
<p>Top ten publishing countries.</p>
Full article ">Figure 6
<p>Top ten most active institutions.</p>
Full article ">Figure 7
<p>Top ten most active Journals.</p>
Full article ">Figure 8
<p>Overlay visualization map of the top 20 frequent author keywords.</p>
Full article ">Figure 9
<p>A Network Visualization of Key Terms in performance Evaluation of e-LMED.</p>
Full article ">Figure 10
<p>An Overlay Visualization of Key Terms in performance Evaluation of e-LMED.</p>
Full article ">Figure 11
<p>Terms co-occurring with performance.</p>
Full article ">Figure 12
<p>Factorial analysis using the Multiple Correspondence Analysis (MCA) method.</p>
Full article ">Figure 13
<p>Dendrogram of hierarchical cluster analysis of keywords displaying the closeness of association between domain keywords.</p>
Full article ">Figure 14
<p>Factorial map of the document with the highest contributions.</p>
Full article ">Figure 15
<p>Factorial map of the most cited documents.</p>
Full article ">Figure 16
<p>Network Visualization map of authors’ collaboration in performance evaluation in e-LMED.</p>
Full article ">Figure 17
<p>Network Visualization map of the largest set of authors’ collaboration in performance evaluation in e-LMED.</p>
Full article ">
23 pages, 5953 KiB  
Article
Hybrid Ensemble Model for Predicting the Strength of FRP Laminates Bonded to the Concrete
by Anas Abdulalem Alabdullh, Rahul Biswas, Jitendra Gudainiyan, Kaffayatullah Khan, Abdullah Hussain Bujbarah, Qasem Ahmed Alabdulwahab, Muhammad Nasir Amin and Mudassir Iqbal
Polymers 2022, 14(17), 3505; https://doi.org/10.3390/polym14173505 - 26 Aug 2022
Cited by 9 | Viewed by 1881
Abstract
The goal of this work was to use a hybrid ensemble machine learning approach to estimate the interfacial bond strength (IFB) of fibre-reinforced polymer laminates (FRPL) bonded to the concrete using the results of a single shear-lap test. A database comprising 136 data [...] Read more.
The goal of this work was to use a hybrid ensemble machine learning approach to estimate the interfacial bond strength (IFB) of fibre-reinforced polymer laminates (FRPL) bonded to the concrete using the results of a single shear-lap test. A database comprising 136 data was used to train and validate six standalone machine learning models, namely, artificial neural network (ANN), extreme machine learning (ELM), the group method of data handling (GMDH), multivariate adaptive regression splines (MARS), least square-support vector machine (LSSVM), and Gaussian process regression (GPR). The hybrid ensemble (HENS) model was subsequently built, employing the combined and trained predicted outputs of the ANN, ELM, GMDH, MARS, LSSVM, and GPR models. In comparison with the standalone models employed in the current investigation, it was observed that the suggested HENS model generated superior predicted accuracy with R2 (training = 0.9783, testing = 0.9287), VAF (training = 97.83, testing = 92.87), RMSE (training = 0.0300, testing = 0.0613), and MAE (training = 0.0212, testing = 0.0443). Using the training and testing dataset to assess the predictive performance of all models for IFB prediction, it was discovered that the HENS model had the greatest predictive accuracy throughout both stages with an R2 of 0.9663. According to the findings of the experiments, the newly developed HENS model has a great deal of promise to be a fresh approach to deal with the overfitting problems of CML models and thus may be utilised to forecast the IFB of FRPL. Full article
(This article belongs to the Special Issue Fiber-Reinforced Polymer Composites in Construction Materials)
Show Figures

Figure 1

Figure 1
<p>Single-lap shear test: (<b>a</b>) FRP externally bonded on concrete; (<b>b</b>) FRP externally bonded on the grooves of concrete (reprinted/adapted with permission from Su et al. [<a href="#B80-polymers-14-03505" class="html-bibr">80</a>]).</p>
Full article ">Figure 2
<p>Example of a MARS model [<a href="#B62-polymers-14-03505" class="html-bibr">62</a>].</p>
Full article ">Figure 3
<p>Pearson correlation with heat map.</p>
Full article ">Figure 4
<p>Sensitivity analysis of input parameters to output parameters.</p>
Full article ">Figure 5
<p>Flowchart of the implementation approach of the HENS Model.</p>
Full article ">Figure 6
<p>Tested vs. predicted graph of training data.</p>
Full article ">Figure 7
<p>Tested vs. predicted graph of testing data.</p>
Full article ">Figure 8
<p>Taylor diagram of the training data.</p>
Full article ">Figure 9
<p>Taylor diagram of the testing data.</p>
Full article ">Figure 10
<p>Accuracy matrix for (<b>a</b>) training, (<b>b</b>) testing, and (<b>c</b>) total datasets.</p>
Full article ">Figure 10 Cont.
<p>Accuracy matrix for (<b>a</b>) training, (<b>b</b>) testing, and (<b>c</b>) total datasets.</p>
Full article ">
15 pages, 3745 KiB  
Article
Computational Analysis of Short Linear Motifs in the Spike Protein of SARS-CoV-2 Variants Provides Possible Clues into the Immune Hijack and Evasion Mechanisms of Omicron Variant
by Anjana Soorajkumar, Ebrahim Alakraf, Mohammed Uddin, Stefan S. Du Plessis, Alawi Alsheikh-Ali and Richard K. Kandasamy
Int. J. Mol. Sci. 2022, 23(15), 8822; https://doi.org/10.3390/ijms23158822 - 8 Aug 2022
Cited by 1 | Viewed by 2685
Abstract
Short linear motifs (SLiMs) are short linear sequences that can mediate protein–protein interaction. Mimicking eukaryotic SLiMs to compete with extra- or intracellular binding partners, or to sequester host proteins is the crucial strategy of viruses to pervert the host system. Evolved proteins in [...] Read more.
Short linear motifs (SLiMs) are short linear sequences that can mediate protein–protein interaction. Mimicking eukaryotic SLiMs to compete with extra- or intracellular binding partners, or to sequester host proteins is the crucial strategy of viruses to pervert the host system. Evolved proteins in viruses facilitate minimal protein–protein interactions that significantly affect intracellular signaling networks. Unfortunately, very little information about SARS-CoV-2 SLiMs is known, especially across SARS-CoV-2 variants. Through the ELM database-based sequence analysis of spike proteins from all the major SARS-CoV-2 variants, we identified four overriding SLiMs in the SARS-CoV-2 Omicron variant, namely, LIG_TRFH_1, LIG_REV1ctd_RIR_1, LIG_CaM_NSCaTE_8, and MOD_LATS_1. These SLiMs are highly likely to interfere with various immune functions, interact with host intracellular proteins, regulate cellular pathways, and lubricate viral infection and transmission. These cellular interactions possibly serve as potential therapeutic targets for these variants, and this approach can be further exploited to combat emerging SARS-CoV-2 variants. Full article
(This article belongs to the Special Issue Host-Pathogen Interaction 4.0)
Show Figures

Figure 1

Figure 1
<p>Domain organization and SLiMs in SARS-CoV-2 variants. (<b>A</b>) Protein domain organization of spike protein. The S-protein contains the N-terminal domain (NTD), receptor binding domain (RBD), subdomain 1 (SD1), subdomain 2 (SD2), fusion peptide (FP), heptad repeat 1 (HR1), central helix (CH), connector domain (CD), heptad repeat 2 (HR2), and cytoplasmic tail (CT). (<b>B</b>) Matrix of mutations on spike protein across SARS-CoV-2 variants. (<b>C</b>) Heatmap summarizing the identified SLiMs across SARS-CoV-2 variants. (<b>D</b>) Venn diagram showing the overlap of SLiMs of the spike protein from Wuhan, Delta, and Omicron variants. (<b>E</b>) A circular dendrogram showing the similarity of the spike protein from SARS-CoV-2 variants.</p>
Full article ">Figure 2
<p>Structure and interaction networks of shelterin-complex-associated proteins. (<b>A</b>) Snapshot of the multiple sequence alignment of spike proteins from Wuhan, Delta, and Omicron, along with the specific mutations in the Omicron that led to the emergence of the novel LIG_TRFH_1 motif. (<b>B</b>) Snapshot of the multiple sequence alignment of spike protein from Omicron variant along with human proteins that contain this specific motif. (<b>C</b>) Domain organization of human TRF1, TRF2, and TIN2. (<b>D</b>) Interaction of the human-telomere-associated proteins. TIN2 bridges TRF1 and TRF2 that bind to the ds telomeric DNA. (<b>E</b>) Omicron SLiM LIG_TRFH_1 interacts with cellular TRF2 and TIN2. (<b>F</b>) LIG_TRFH_1 interaction with shelterin proteins protects the viral terminal repeats. Other proteins involved in the protective complex must be identified. TR-terminal repeats.</p>
Full article ">Figure 3
<p>Representation of DNA damage tolerance pathway. (<b>A</b>,<b>B</b>) Snapshot of the multiple sequence alignment of spike proteins from Wuhan, Delta, and Omicron along with the specific mutations in the Omicron that lead to the emergence of SLiMs. (<b>C</b>) Schematic illustration of the domain structure of human Rev1. (<b>D</b>) Interactions of human Y-family polymerases in TLS. (<b>E</b>) Omicron SLiMs LIG_REV1ctd_RIR_1 and LIG_PCNA_TLS_4 interact with human REV1. (<b>F</b>) LIG_REV1ctd_RIR_1 and LIG_PCNA_TLS_4 motif involvement in the viral TLS.</p>
Full article ">Figure 4
<p>The interaction of NSCaTE and IQ motifs in a Ca<sup>2+</sup>/CaM-mediated manner. Snapshot of the multiple sequence alignment of spike proteins from (<b>A</b>) SARS-CoV-2 (Wuhan), SARS1, and MERS1 viruses; and (<b>B</b>) Wuhan, Delta, and Omicron along with the specific mutations in the Omicron that lead to the emergence of SLiMs. (<b>C</b>) Ca<sup>2+</sup> influx facilitates motif interactions. In resting state, motifs remain unbound, NSCaTE in the N-terminus, and the IQ motif in the C-terminus. Upon membrane depolarization and Ca<sup>2+</sup> influx, a Ca<sup>2+</sup>/CaM-complex-mediated interaction of both motifs occurs in the Cav1 channel. (<b>D</b>) A cartoon representation of the involvement of increased LIG_CaM_NSCaTE_8 SLiM-mediated transmissibility in the SARS-CoV-2 Omicron variant. The spike glycoprotein on SARS-CoV-2 interacts with ACE2 to enter the host cells. Viral entry results in an intracellular hike on the Ca<sup>2+</sup> level and hence the Ca<sup>2+</sup>/CaM complex in the cells. Ca<sup>2+</sup>/CaM complex-mediated ACE2 catalytic ectodomain shedding by ADAM-17 generates the soluble form of ACE2. SARS-CoV-2 can bind to the soluble ACE2, as it contains the viral binding site, but viral neutralization occurs without an intracellular environment and cannot duplicate. When the Omicron variant enters the cells, Ca<sup>2+</sup>/CaM-mediated binding transpires between the unique Omicron SLiM LIG_CaM_NSCaTE_8 and the LIG_CaM_IQ_9. This process hinders ACE2 ectodomain shedding due to the lack of Ca<sup>2+</sup>/CaM complex availability for the CaM binding site in the ACE2 cytoplasmic receptor. As a result, more active full-length ACE2 is expressed on the surface for viral binding.</p>
Full article ">Figure 5
<p>Schematic representation of the Hippo signaling pathway and Omicron MOD_LATS_1 intervention. (<b>A</b>) Snapshot of the multiple sequence alignment of spike proteins from Wuhan, Delta, and Omicron along with the specific mutations in the Omicron that lead to the emergence of the SLiMs. (<b>B</b>) Modulation in the Hippo signaling pathway during Omicron infection. When the Hippo signaling pathway is active/on, YAP/TAZ proteins become phosphorylated by LATS1/2 kinases and remain in inactive form. However, during Omicron infection, cellular LATS1/2 kinases phosphorylate viral MOD_LATS_1 leaving active cytoplasmic YAP/TAZ, which can negatively regulate immune response and facilitate Omicron survival.</p>
Full article ">Figure 6
<p>Schematic representation of Omicron spike protein organization and amino acid mutations. Omicron mutations are shown in a primary structure of SARS-CoV-2 S-protein. Amino acid mutations in SARS-CoV-2 Omicron spike proteins are A67V, Del69-70, T95I, Del142-144, Y145D, Del211, L212I, R214Insertion, G339D, S371L, S373P, S375F, K417N, N440K, G446S, S477N, T478K, E484A, Q493R, G496S, Q498R, N501Y, Y505H, T547K, D614G, H655Y, N679K, P681H, N764K, D796Y, N856K, Q954H, N969K, and L981F. Selected SLiMs introduced due to the mutation in Omicron variants are marked at the bottom of the domain map.</p>
Full article ">
23 pages, 3243 KiB  
Article
Comparative Analysis of Transcriptomes of Ophiostoma novo-ulmi ssp. americana Colonizing Resistant or Sensitive Genotypes of American Elm
by Martha Nigg, Thais C. de Oliveira, Jorge L. Sarmiento-Villamil, Paul Y. de la Bastide, Will E. Hintz, Sherif M. Sherif, Mukund Shukla, Louis Bernier and Praveen K. Saxena
J. Fungi 2022, 8(6), 637; https://doi.org/10.3390/jof8060637 - 16 Jun 2022
Cited by 6 | Viewed by 3621
Abstract
The Ascomycete Ophiostoma novo-ulmi threatens elm populations worldwide. The molecular mechanisms underlying its pathogenicity and virulence are still largely uncharacterized. As part of a collaborative study of the O. novo-ulmi-elm interactome, we analyzed the O. novo-ulmi ssp. americana transcriptomes obtained by deep [...] Read more.
The Ascomycete Ophiostoma novo-ulmi threatens elm populations worldwide. The molecular mechanisms underlying its pathogenicity and virulence are still largely uncharacterized. As part of a collaborative study of the O. novo-ulmi-elm interactome, we analyzed the O. novo-ulmi ssp. americana transcriptomes obtained by deep sequencing of messenger RNAs recovered from Ulmus americana saplings from one resistant (Valley Forge, VF) and one susceptible (S) elm genotypes at 0 and 96 h post-inoculation (hpi). Transcripts were identified for 6424 of the 8640 protein-coding genes annotated in the O. novo-ulmi nuclear genome. A total of 1439 genes expressed in planta had orthologs in the PHI-base curated database of genes involved in host-pathogen interactions, whereas 472 genes were considered differentially expressed (DEG) in S elms (370 genes) and VF elms (102 genes) at 96 hpi. Gene ontology (GO) terms for processes and activities associated with transport and transmembrane transport accounted for half (27/55) of GO terms that were significantly enriched in fungal genes upregulated in S elms, whereas the 22 GO terms enriched in genes overexpressed in VF elms included nine GO terms associated with metabolism, catabolism and transport of carbohydrates. Weighted gene co-expression network analysis identified three modules that were significantly associated with higher gene expression in S elms. The three modules accounted for 727 genes expressed in planta and included 103 DEGs upregulated in S elms. Knockdown- and knockout mutants were obtained for eight O. novo-ulmi genes. Although mutants remained virulent towards U. americana saplings, we identified a large repertoire of additional candidate O. novo-ulmi pathogenicity genes for functional validation by loss-of-function approaches. Full article
(This article belongs to the Special Issue Dutch Elm Disease in the 21st Century)
Show Figures

Figure 1

Figure 1
<p>Differentially expressed <span class="html-italic">Ophiostoma novo-ulmi</span> genes during interaction with <span class="html-italic">Ulmus americana</span>. (<b>A</b>) MA plot of the 6424 <span class="html-italic">O. novo-ulmi</span> genes that were detected at 96 hpi showing genes that were overexpressed in susceptible (S) and resistant (VF) elm genotypes. Read counts were normalized using DESeq2 in R [<a href="#B56-jof-08-00637" class="html-bibr">56</a>] with log<sub>2</sub>FC &gt; 1 and scatter plot visualization. (<b>B</b>) Gene ontology (GO) terms for biological processes that were significantly enriched in <span class="html-italic">O. novo-ulmi</span> colonizing susceptible (S) or resistant (VF) elm genotypes. The top 10 processes are shown in the case of interactions with S elm (out of 39 GO terms for biological processes), whereas all enriched terms for processes are shown in the case of interactions with VF elm.</p>
Full article ">Figure 2
<p>Weighted gene co-expression network analysis (WGCNA) of the fungal component of the <span class="html-italic">Ulmus americana</span>-<span class="html-italic">Ophiostoma novo-ulmi</span> ssp. <span class="html-italic">Americana</span> interactome. (<b>A</b>) Cluster dendrogram showing the genes (branches) and co-expressed modules (colors); genes were clustered in 29 modules according to 1-TOM soft threshold. (<b>B</b>) Number of eigengenes per module. (<b>C</b>) Module traits in <span class="html-italic">O. novo-ulmi</span> in planta. The colour scale (red-blue) for relationships between module eigengenes (rows) and treatments (columns) represents the strength of the correlation (1 to −1). Modules Darkred (<span class="html-italic">p</span> &lt; 0.01), Green (<span class="html-italic">p</span> &lt; 0.01) and Purple (<span class="html-italic">p</span> &lt; 0.05) were associated with higher gene expression in susceptible (S) compared to resistant (VF) elm.</p>
Full article ">Figure 3
<p>Annotation of genes found in weighted gene co-expression network analysis modules Darkred, Green and Purple that were associated with higher <span class="html-italic">Ohiostoma novo-ulmi</span> ssp. <span class="html-italic">Americana</span> gene expression in susceptible <span class="html-italic">Ulmus americana</span>. (<b>A</b>) Identity and expression level of genes encoding a protein with a signal peptide. Number of reads is shown for each replicate. (<b>B</b>) CAZY classes. (<b>C</b>) PHI-base classes. (<b>D</b>) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway classes.</p>
Full article ">Figure 4
<p>Interaction networks for selected genes in weighted-gene co-expression network analysis modules Darkred (<b>A</b>), Green (<b>B</b>) and Purple (<b>C</b>) that were associated with higher <span class="html-italic">Ohiostoma novo-ulmi</span> ssp. <span class="html-italic">americana</span> gene expression in susceptible (S) <span class="html-italic">Ulmus americana</span>. Interactions are shown for genes associated with KEGG pathway classes, along with genes predicted to encode a secreted protein with a signal peptide (light blue), and genes for which knockdown- or knockout mutants (triangles) were obtained and tested. Genes upregulated in S elms are identified by a star. PHI-base identifier is shown for genes with orthologs in the PHI-base curated database of genes involved in host-pathogen interactions. The size of the gene symbol (circle or triangle) is proportional to the number of interactions with other genes in the module.</p>
Full article ">
22 pages, 4183 KiB  
Article
Comparative Analysis of Structural Features in SLiMs from Eukaryotes, Bacteria, and Viruses with Importance for Host-Pathogen Interactions
by Heidy Elkhaligy, Christian A. Balbin and Jessica Siltberg-Liberles
Pathogens 2022, 11(5), 583; https://doi.org/10.3390/pathogens11050583 - 15 May 2022
Cited by 2 | Viewed by 2721
Abstract
Protein-protein interactions drive functions in eukaryotes that can be described by short linear motifs (SLiMs). Conservation of SLiMs help illuminate functional SLiMs in eukaryotic protein families. However, the simplicity of eukaryotic SLiMs makes them appear by chance due to mutational processes not only [...] Read more.
Protein-protein interactions drive functions in eukaryotes that can be described by short linear motifs (SLiMs). Conservation of SLiMs help illuminate functional SLiMs in eukaryotic protein families. However, the simplicity of eukaryotic SLiMs makes them appear by chance due to mutational processes not only in eukaryotes but also in pathogenic bacteria and viruses. Further, functional eukaryotic SLiMs are often found in disordered regions. Although proteomes from pathogenic bacteria and viruses have less disorder than eukaryotic proteomes, their proteins can successfully mimic eukaryotic SLiMs and disrupt host cellular function. Identifying important SLiMs in pathogens is difficult but essential for understanding potential host-pathogen interactions. We performed a comparative analysis of structural features for experimentally verified SLiMs from the Eukaryotic Linear Motif (ELM) database across viruses, bacteria, and eukaryotes. Our results revealed that many viral SLiMs and specific motifs found across viruses and eukaryotes, such as some glycosylation motifs, have less disorder. Analyzing the disorder and coil properties of equivalent SLiMs from pathogens and eukaryotes revealed that some motifs are more structured in pathogens than their eukaryotic counterparts and vice versa. These results support a varying mechanism of interaction between pathogens and their eukaryotic hosts for some of the same motifs. Full article
(This article belongs to the Special Issue Computational Biology Applied to Host-Pathogen Interactions)
Show Figures

Figure 1

Figure 1
<p>The SLiM dataset composition by taxonomy and functionality. The percentage of SLiMs per taxonomic group and taxonomic subgroup; eukaryotes and its subgroups (grey), viruses and its subgroups (blue), and bacteria (green) based on all SLiMs (<b>A</b>). The percentage of SLiMs is colored by functional type in each taxonomic group (<b>B</b>). For further information, see <a href="#app1-pathogens-11-00583" class="html-app">Table S1</a>.</p>
Full article ">Figure 2
<p>Predicted properties per instance across taxonomic groups. The predicted percentage per instance; IUPRED2A long disorder based on 0.5 cutoff (<b>A</b>) and 0.4 cutoff (<b>B</b>), IUPRED2A short disorder based on 0.5 cutoff (<b>C</b>) and 0.4 cutoff (<b>D</b>), NetSurfP 2.0 accessibility based on 0.25 cutoff (<b>E</b>), and NetSurfP 2.0 prediction of coil based on three state analysis (<b>F</b>). For further information, see <a href="#app1-pathogens-11-00583" class="html-app">Table S1</a>.</p>
Full article ">Figure 3
<p>Distribution of MIDS values. Boxplots for the distribution of long IUPRED2A MIDS of all SLiMs per motif type colored as shown by legend (<b>A</b>). Boxplots for long IUPRED2A MIDS distribution of all SLiMs in each taxonomic group (bacteria (green), viruses (blue), eukaryotes (grey)) classified based on their ELM type (<b>B</b>). Boxplots for the distribution of long IUPRED2A MIDS of all SLiMs per motif type colored as shown by legend (<b>C</b>). Boxplots for long IUPRED2A MIDS distribution of all SLiMs in each taxonomic group, colored as in (<b>B</b>), classified based on their ELM type (<b>D</b>). Hypothesis testing with Mann–Whitney test with simple Bonferroni correction was performed and significant adjusted <span class="html-italic">p</span>-values in (<b>A</b>,<b>B</b>) are shown as brackets between groups (No asterisk for adjusted <span class="html-italic">p</span>-values between 0.05 and &lt;0.01, * for adjusted <span class="html-italic">p</span>-value ≤ 0.01, ** for ≤1 × 10<sup>−3</sup>, and *** for ≤1 × 10<sup>−4</sup>). The sample size per each tested group and adjusted <span class="html-italic">p</span>-values can be found in <a href="#app1-pathogens-11-00583" class="html-app">Table S1</a>. The percentage of SLiMs by long IUPED2A MIDS range in different taxonomic groups colored by ELM type (<b>E</b>–<b>G</b>). The percentage of SLiMs by short IUPED2A MIDS range in different taxonomic groups colored by ELM type (<b>H</b>–<b>J</b>), colored as in (<b>A</b>). For more information, see <a href="#app1-pathogens-11-00583" class="html-app">Tables S1 and S2</a>.</p>
Full article ">Figure 4
<p><b>Distribution of MCCS values.</b> Boxplots for the distribution of MCCS of all SLiMs per motif type colored as shown by legend (<b>A</b>). Boxplots for MCCS distribution of all SLiMs in each taxonomic group (bacteria in green, viruses in blue, and eukaryotes in grey) classified based on their ELM type (<b>B</b>). Hypothesis testing with Mann–Whitney test with simple Bonferroni correction was performed and significant adjusted <span class="html-italic">p</span>-values in (<b>A</b>,<b>B</b>) are shown as brackets between groups (No asterisk for adjusted <span class="html-italic">p</span>-values between 0.05 to &lt;0.01, * for adjusted <span class="html-italic">p</span>-value ≤ 0.01, and *** for ≤1 × 10<sup>−4</sup>). The sample size per each tested group and adjusted <span class="html-italic">p</span>-values can be found in <a href="#app1-pathogens-11-00583" class="html-app">Table S1</a>. The percentage of SLiMs by MCCS range in different taxonomic groups colored by ELM type (<b>C</b>–<b>E</b>) colored as in (<b>A</b>). For more information, see <a href="#app1-pathogens-11-00583" class="html-app">Tables S1 and S2</a>.</p>
Full article ">Figure 5
<p>Disorder and coil confidence profiles of proteins containing SLiMs and the density curve of MIDS and MCCS of SLiMs per taxonomic group. The flanking regions of 100 residues around SLiMs using long IUPRED2A disorder score per taxonomic group and the 95% confidence interval of the mean (<b>A</b>). SLiMs long IUPRED2A MIDS density distribution plot of the SLiMs per taxonomic group (<b>B</b>). The flanking regions of 100 residues around SLiMs using short IUPRED2A disorder score per taxonomic group and the 95% confidence interval of the mean (<b>C</b>). SLiMs short IUPRED2A MIDS density distribution plot of the SLiMs per taxonomic group (<b>D</b>). The flanking regions of 100 residues around SLiMs coil confidence score per taxonomic group and the 95% confidence interval of the mean (<b>E</b>). SLiMs MCCS density distribution plot of the SLiMs per taxonomic group (<b>F</b>). For further information, see <a href="#app1-pathogens-11-00583" class="html-app">Table S3</a>.</p>
Full article ">Figure 6
<p>Scatter plot for the MIDS and MCCS means of the shared SLiMs between different groups. Long disorder MIDS means scatter plot and Spearman correlation with the <span class="html-italic">p</span>-value for shared SLiMs between eukaryotes vs. bacteria (<b>A</b>) and eukaryotes vs. viruses (<b>B</b>). Short disorder MIDS means scatter plot and Spearman correlation with the <span class="html-italic">p</span>-value for shared SLiMs between eukaryotes vs. bacteria (<b>C</b>) and eukaryotes vs. viruses (<b>D</b>). MCCS means scatter plot and Spearman correlation with the <span class="html-italic">p</span>-value for shared SLiMs between eukaryotes vs. bacteria (<b>E</b>) and eukaryotes vs. viruses (<b>F</b>). For detailed information about the number of instances, long/short mMIDS and mMCCS of all instances per motif, long/short MIDS and MCCS per instance, and the individual amino acid scores of disorder and coil confidence per instance, see <a href="#app1-pathogens-11-00583" class="html-app">Table S4</a>.</p>
Full article ">Figure 7
<p>Disorder score and coil confidence distributions in viruses and eukaryotes for the MOD_N-GLC_1 motif. Boxplots and swarm plot distribution for SLiMs long IUPRED2A MIDS (<b>A</b>), short IUPRED2A MIDS (<b>B</b>), MCCS (<b>C</b>), the individual long IUPRED2A disorder scores per residue for SLiMs (<b>D</b>), the individual short IUPRED2A disorder scores per residue for SLiMs (<b>E</b>), and the individual coil confidence scores per residue for SLiMs (<b>F</b>).</p>
Full article ">Figure 8
<p>The glycosylated MOD_N-GLC_1 site in West Nile virus envelope protein. West Nile Virus envelope protein (beige) (PDB ID: 2HG0) rendered as a transparent surface. A closer view of the local helical structure of the MOD_N-GLC_1 motif (magenta). The glycosylated asparagine residue (blue) and glycan group (cyan) are shown as sticks.</p>
Full article ">Figure 9
<p>Phylogenetic tree of West Nile Virus (WNV) envelope protein illustrating the evolution of structural properties of a MOD_N-GLC_1 motif. The tree, rooted by the outgroup Yellow Fever virus (YFV)), shows WNV in green and Zika virus (ZIKV), Dengue virus 2 (DENV2), and Japanese Encephalitis Virus (JEV) that have been shown to be glycosylated in this position but that are not in the ELM database in blue. The tree is shown next to an excerpt from the multiple sequence alignment with the MOD_N-GLC_1 motif pattern highlighted in black, followed by the same alignment excerpt colored by the accessibility and secondary structure of the residues (<b>A</b>) and by disorder using both 0.5 and 0.4 cutoff values for long IUPRED2A and short IUPRED2A disorder, with the location of the WNV MOD_N-GLC_1 motif shown by the black box (<b>B</b>). For further details, see <a href="#app1-pathogens-11-00583" class="html-app">Figure S5</a>.</p>
Full article ">Figure 10
<p>Disorder score and coil confidence distributions in viruses and eukaryotes for the LIG_Rb_LxCxE_1 motif. Boxplots and swarm plot distribution for SLiMs long IUPRED2A MIDS (<b>A</b>), short IUPRED2A MIDS (<b>B</b>), MCCS (<b>C</b>), individual long IUPRED2A disorder scores per residue for SLiMs (<b>D</b>), individual short IUPRED2A disorder scores per residue for SLiMs (<b>E</b>), and individual coil confidence scores per residue for SLiMs (<b>F</b>).</p>
Full article ">Figure 11
<p>LIG_Rb_LxCxE_1 motif segment from Simian V40 (large T antigen protein) and human papillomaviruses (E7) proteins in a bound state with retinoblastoma protein. The complete structures from PDB ID: 1GH6 and PDB ID: 1GUX are aligned, and a closer view of the LxCxE binding site is shown. Retinoblastoma protein (beige and cyan) is rendered as a cartoon. Large T antigen protein is shown as a cartoon (dark pink). The E7 of the human papillomavirus motif segment is shown as ribbon (brown). The LxCxE motif in both proteins is shown as sticks. The structural alignment of the entire two structures was performed in PyMOL (PyMOL Molecular Graphics System, Version 4.6).</p>
Full article ">
14 pages, 1329 KiB  
Article
MSPEDTI: Prediction of Drug–Target Interactions via Molecular Structure with Protein Evolutionary Information
by Lei Wang, Leon Wong, Zhan-Heng Chen, Jing Hu, Xiao-Fei Sun, Yang Li and Zhu-Hong You
Biology 2022, 11(5), 740; https://doi.org/10.3390/biology11050740 - 13 May 2022
Cited by 7 | Viewed by 4332
Abstract
The key to new drug discovery and development is first and foremost the search for molecular targets of drugs, thus advancing drug discovery and drug repositioning. However, traditional drug–target interactions (DTIs) is a costly, lengthy, high-risk, and low-success-rate system project. Therefore, more and [...] Read more.
The key to new drug discovery and development is first and foremost the search for molecular targets of drugs, thus advancing drug discovery and drug repositioning. However, traditional drug–target interactions (DTIs) is a costly, lengthy, high-risk, and low-success-rate system project. Therefore, more and more pharmaceutical companies are trying to use computational technologies to screen existing drug molecules and mine new drugs, leading to accelerating new drug development. In the current study, we designed a deep learning computational model MSPEDTI based on Molecular Structure and Protein Evolutionary to predict the potential DTIs. The model first fuses protein evolutionary information and drug structure information, then a deep learning convolutional neural network (CNN) to mine its hidden features, and finally accurately predicts the associated DTIs by extreme learning machine (ELM). In cross-validation experiments, MSPEDTI achieved 94.19%, 90.95%, 87.95%, and 86.11% prediction accuracy in the gold-standard datasets enzymes, ion channels, G-protein-coupled receptors (GPCRs), and nuclear receptors, respectively. MSPEDTI showed its competitive ability in ablation experiments and comparison with previous excellent methods. Additionally, 7 of 10 potential DTIs predicted by MSPEDTI were substantiated by the classical database. These excellent outcomes demonstrate the ability of MSPEDTI to provide reliable drug candidate targets and strongly facilitate the development of drug repositioning and drug development. Full article
(This article belongs to the Special Issue Intelligent Computing in Biology and Medicine)
Show Figures

Figure 1

Figure 1
<p>Schematic diagram of the structure of CNN.</p>
Full article ">Figure 2
<p>ROC of 5CV mapped by MSPEDTI on enzyme dataset.</p>
Full article ">Figure 3
<p>ROC of 5CV mapped by MSPEDTI on ion channel dataset.</p>
Full article ">Figure 4
<p>ROC of 5CV mapped by MSPEDTI on GPCR dataset.</p>
Full article ">Figure 5
<p>ROC of 5CV mapped by MSPEDTI on nuclear receptor dataset.</p>
Full article ">Figure 6
<p>ROC curves plotted by the 2DPCA descriptor model on ion channel.</p>
Full article ">Figure 7
<p>ROC curves plotted by the SVM classifier model on ion channel.</p>
Full article ">
20 pages, 4242 KiB  
Review
Dynamic, but Not Necessarily Disordered, Human-Virus Interactions Mediated through SLiMs in Viral Proteins
by Heidy Elkhaligy, Christian A. Balbin, Jessica L. Gonzalez, Teresa Liberatore and Jessica Siltberg-Liberles
Viruses 2021, 13(12), 2369; https://doi.org/10.3390/v13122369 - 26 Nov 2021
Cited by 10 | Viewed by 3478
Abstract
Most viruses have small genomes that encode proteins needed to perform essential enzymatic functions. Across virus families, primary enzyme functions are under functional constraint; however, secondary functions mediated by exposed protein surfaces that promote interactions with the host proteins may be less constrained. [...] Read more.
Most viruses have small genomes that encode proteins needed to perform essential enzymatic functions. Across virus families, primary enzyme functions are under functional constraint; however, secondary functions mediated by exposed protein surfaces that promote interactions with the host proteins may be less constrained. Viruses often form transient interactions with host proteins through conformationally flexible interfaces. Exposed flexible amino acid residues are known to evolve rapidly suggesting that secondary functions may generate diverse interaction potentials between viruses within the same viral family. One mechanism of interaction is viral mimicry through short linear motifs (SLiMs) that act as functional signatures in host proteins. Viral SLiMs display specific patterns of adjacent amino acids that resemble their host SLiMs and may occur by chance numerous times in viral proteins due to mutational and selective processes. Through mimicry of SLiMs in the host cell proteome, viruses can interfere with the protein interaction network of the host and utilize the host-cell machinery to their benefit. The overlap between rapidly evolving protein regions and the location of functionally critical SLiMs suggest that these motifs and their functional potential may be rapidly rewired causing variation in pathogenicity, infectivity, and virulence of related viruses. The following review provides an overview of known viral SLiMs with select examples of their role in the life cycle of a virus, and a discussion of the structural properties of experimentally validated SLiMs highlighting that a large portion of known viral SLiMs are devoid of predicted intrinsic disorder based on the viral SLiMs from the ELM database. Full article
(This article belongs to the Special Issue Host Cell-Virus Interaction)
Show Figures

Figure 1

Figure 1
<p>Predicted structural features of 260 viral SLiMs from the ELM database. The percentage of viral motifs with a certain disorder content as inferred from IUPRED prediction using a cutoff of (<b>a</b>) 0.5 and (<b>b</b>) 0.4. (<b>c</b>) The percentage of viral motifs with a certain Mean IUPRED Disorder Score (MIDS). The percentage of viral motifs with a certain (<b>d</b>) secondary structure (coil) and (<b>e</b>) surface accessibility content as inferred from NetSurfP-2.0 prediction. The percentages shown are approximate; rounded to the nearest whole number for a, b, d, and e, and to the nearest tenth for c. See also <a href="#app1-viruses-13-02369" class="html-app">Table S1</a>.</p>
Full article ">Figure 2
<p>The general lytic virus life cycle inside the cells. (1) The virion attaches to the cell surface receptors. (2) The penetration of the virus through endocytosis to the infected cell. (3) The replicated genome and translated viral proteins inside the cell. (4) The newly assembled viruses inside the cell. (5) The cell lysis and release of new viruses from the infected cell. Created with <a href="http://BioRender.com" target="_blank">BioRender.com</a> (accessed on 30 October 2021).</p>
Full article ">Figure 3
<p>The furin cleavage site in the envelope glycoprotein from HIV. Sequences were identified with BLAST using the envelope protein (accession: NP_057856.1) from HIV-1 as query. Sequence names shown in red represents true positive instances from the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>]. The multiple sequence alignment (MSA) was built with MAFFT+L-INS-i [<a href="#B57-viruses-13-02369" class="html-bibr">57</a>] in Jalview [<a href="#B58-viruses-13-02369" class="html-bibr">58</a>]. The regular expression pattern R.[RK]R. from motif CLV_PCSK_FUR_1 in the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>] was identified using Find in Jalview, shown in black with white text. The region shown under Sequence shows the amino acids that corresponds to the true positive motif from ENV_HIV1 plus one additional site on each side. The three additional heatmaps display the same region of the alignment colored by property. The heatmap for Disorder propensity displays disordered (magenta) or ordered (purple) residues based on IUPRED prediction with cutoff = 0.4 [<a href="#B35-viruses-13-02369" class="html-bibr">35</a>,<a href="#B36-viruses-13-02369" class="html-bibr">36</a>,<a href="#B59-viruses-13-02369" class="html-bibr">59</a>]. Heatmaps for (1) Surface accessibility displays surface exposed (magenta) and buried (white) residues and (2) Secondary structure displays coil (orange) and secondary structure (helix: blue, strand: magenta) based on NetSurfP-2.0 predictions.</p>
Full article ">Figure 4
<p>The G3BP binding motif has been verified in the nsp3 protein from Chikungunya virus and Semliki Forest virus from Alphaviruses. Sequences were identified with BLAST using residues 1700–2000 from nsp3 (accession: Q5XXP4) from Chikungunya virus as query. Sequence names shown in red represents true positive instances from the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>]. The multiple sequence alignment was built with MAFFT+L-INS-i [<a href="#B57-viruses-13-02369" class="html-bibr">57</a>] in Jalview [<a href="#B58-viruses-13-02369" class="html-bibr">58</a>]. The regular expression pattern [FYLIMV].FG[DES]F from motif LIG_G3BP_FGDF_1 in the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>] was identified using Find in Jalview, shown in black with white text. The region shown under Sequence shows the amino acids that corresponds to the true positive motifs from Chikungunya virus and Semliki Forest virus, the connecting amino acids, plus one additional site on each side. The MSA and heatmaps for Disorder, Surface, and Structure are colored as in <a href="#viruses-13-02369-f003" class="html-fig">Figure 3</a>.</p>
Full article ">Figure 5
<p>The pLxIS site in nsp1 from Simian rotavirus. Sequences were identified with BLAST using full-length nsp1 from Simian rotavirus (accession: AFY98633.1) as query. Sequence names shown in red represents true positive instances from the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>]. The multiple sequence alignment was built with MAFFT+L-INS-i [<a href="#B57-viruses-13-02369" class="html-bibr">57</a>] in Jalview [<a href="#B58-viruses-13-02369" class="html-bibr">58</a>]. The regular expression pattern [VILPF].{1,3}L.I(S) from motif LIG_IRF3_LxIS_1 in the ELM database was identified using Find in Jalview, shown in black with white text. The region shown under Sequence shows the amino acids that corresponds to the true positive motif from Simian rotavirus plus one additional site on each side. The MSA and heatmaps for Disorder, Surface, and Structure are colored as in <a href="#viruses-13-02369-f003" class="html-fig">Figure 3</a>.</p>
Full article ">Figure 6
<p>The PDZ domain binding motif in the E6 protein from HPV16 and HPV18. Sequences were identified with BLAST using protein E6 from HPV18 (accession: P06463.1) as query. Sequence names shown in red represents true positive instances from the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>]. The multiple sequence alignment (MSA) was built with MAFFT+L-INS-i [<a href="#B57-viruses-13-02369" class="html-bibr">57</a>] in Jalview [<a href="#B58-viruses-13-02369" class="html-bibr">58</a>]. The regular expression pattern …[ST].[ACVILF]<span>$</span> from motif LIG_PDZ_Class_1 in the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>] was identified using Find in Jalview, shown in black with white text. The region shown under Sequence shows the amino acids that corresponds to the true positive motif from HPV16 and HPV18 plus one additional site on each side. The MSA and heatmaps for Disorder, Surface, and Structure are colored as in <a href="#viruses-13-02369-f003" class="html-fig">Figure 3</a>.</p>
Full article ">Figure 7
<p>The PPxY motif in the matrix protein VP40 from Ebola virus. Sequences were identified with BLAST using full-length VP40 from Ebola virus (accession: Q05128) as query against the refseq_protein and nr databases. Sequence names shown in red represents true positive instances from the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>]. The multiple sequence alignment was built with MAFFT+L-INS-i [<a href="#B57-viruses-13-02369" class="html-bibr">57</a>] in Jalview [<a href="#B58-viruses-13-02369" class="html-bibr">58</a>]. The regular expression pattern PP.Y from motif LIG_WW_1 in the ELM database [<a href="#B27-viruses-13-02369" class="html-bibr">27</a>] was identified using Find in Jalview, shown in black with white text. The region shown under Sequence corresponds to the true positive motif from Zaire Ebola virus and Marburg marburg virus plus one additional site on each side. It should be noted that query protein Q05128 Uniprot ID is identical to protein NP_066245.1 used in the multiple sequence alignment.</p>
Full article ">Figure 8
<p>Cellular context. Subcellular localization of SARS-CoV-2 proteins (circles) in human cells based on experimental data (thick border: multiple sources; dotted border: [<a href="#B127-viruses-13-02369" class="html-bibr">127</a>]; thin black border: [<a href="#B128-viruses-13-02369" class="html-bibr">128</a>]; white border: [<a href="#B129-viruses-13-02369" class="html-bibr">129</a>,<a href="#B130-viruses-13-02369" class="html-bibr">130</a>,<a href="#B131-viruses-13-02369" class="html-bibr">131</a>]). (<b>a</b>). Each protein is colored as in the SARS-CoV-2 proteome (<b>b</b>). Proteins that form complexes are colored similarly; nsp 3/4/6, nsp 7/8/12, nsp 10/14. SARS-CoV-2 proteins localize to the following organelles: lysosome (nsp2, orf3a, and orf7b), endosome (orf3a and orf6), plasma membrane (envelope (E), membrane (M), spike (S), and orf3a), Golgi apparatus (E, M, S, nsp5, nsp15, orf6, orf7a, and orf7b), endoplasmic reticulum (E, M, S, nsp6-10, nsp14, orf6, orf7b, orf8, and orf10), nucleolus (E, nsp1, nsp3, nsp5-7, nsp9-10, nsp12-16 and orf9a-9b), punctate cytoplasm (M, nsp1, nsp2, nsp5, nsp7-10, nsp12-16, orf3a, and orf6), and diffuse cytoplasm (E, M, nucleocapsid (N), S, nsp1-16, nsp10, nsp12-16, orf3a-3b, orf6, orf7a-7b, orf8, orf9a-9b, and orf10). Created with <a href="http://BioRender.com" target="_blank">BioRender.com</a> (accessed on 30 October 2021).</p>
Full article ">
Back to TopTop