Predicting the effects of mutations on protein solubility using graph convolution network and protein language model representation

Solubility is one of the most important properties of protein. Protein solubility can be greatly changed by single amino acid mutations and the reduced protein solubility could lead to diseases. Since experimental methods to determine solubility are time-consuming and expensive, in-silico methods have been developed to predict the protein solubility changes caused by mutations mostly through protein evolution information. However, these methods are slow since it takes long time to obtain evolution information through multiple sequence alignment. In addition, these methods are of low performance because they do not fully utilize protein 3D structures due to a lack of experimental structures for most proteins. Here, we proposed a sequence-based method DeepMutSol to predict solubility change from residual mutations based on the Graph Convolutional Neural Network (GCN), where the protein graph was initiated according to predicted protein structure from Alphafold2, and the nodes (residues) were represented by protein language embeddings. To circumvent the small data of solubility changes, we further pretrained the model over absolute protein solubility. DeepMutSol was shown to outperform state-of-the-art methods in benchmark tests. In addition, we applied the method to clinically relevant genes from the ClinVar database and the predicted solubility changes were shown able to separate pathogenic mutations. All of the data sets and the source code are available at https://github.com/biomed-AI/DeepMutSol.

Personalized Driver Gene Prediction Using Graph Convolutional Networks with Conditional Random Fields

Cancer is a complex and evolutionary disease mainly driven by the accumulation of genetic variations in genes. Identifying cancer driver genes is important. However, most related studies have focused on the population level. Cancer is a disease with high heterogeneity. Thus, the discovery of driver genes at the individual level is becoming more valuable but is a great challenge. Although there have been some computational methods proposed to tackle this challenge, few can cover all patient samples well, and there is still room for performance improvement. In this study, to identify individual-level driver genes more efficiently, we propose the PDGCN method. PDGCN integrates multiple types of data features, including mutation, expression, methylation, copy number data, and system-level gene features, along with network structural features extracted using Node2vec in order to construct a sample-gene interaction network. Prediction is performed using a graphical convolutional neural network model with a conditional random field layer, which is able to better combine the network structural features with biological attribute features. Experiments on the ACC (Adrenocortical Cancer) and KICH (Kidney Chromophobe) datasets from TCGA (The Cancer Genome Atlas) demonstrated that the method performs better compared to other similar methods. It can identify not only frequently mutated driver genes, but also rare candidate driver genes and novel biomarker genes. The results of the survival and enrichment analyses of these detected genes demonstrate that the method can identify important driver genes at the individual level.

Utilizing graph convolutional networks for identification of mild cognitive impairment from single modal fMRI data: a multiconnection pattern combination approach

Mild cognitive impairment plays a crucial role in predicting the early progression of Alzheimer's disease, and it can be used as an important indicator of the disease progression. Currently, numerous studies have focused on utilizing the functional brain network as a novel biomarker for mild cognitive impairment diagnosis. In this context, we employed a graph convolutional neural network to automatically extract functional brain network features, eliminating the need for manual feature extraction, to improve the mild cognitive impairment diagnosis performance. However, previous graph convolutional neural network approaches have primarily concentrated on single modes of brain connectivity, leading to a failure to leverage the potential complementary information offered by diverse connectivity patterns and limiting their efficacy. To address this limitation, we introduce a novel method called the graph convolutional neural network with multimodel connectivity, which integrates multimode connectivity for the identification of mild cognitive impairment using fMRI data and evaluates the graph convolutional neural network with multimodel connectivity approach through a mild cognitive impairment diagnostic task on the Alzheimer's Disease Neuroimaging Initiative dataset. Overall, our experimental results show the superiority of the proposed graph convolutional neural network with multimodel connectivity approach, achieving an accuracy rate of 92.2% and an area under the Receiver Operating Characteristic (ROC) curve of 0.988.

EEG-based high-performance depression state recognition

Depression is a global disease that is harmful to people. Traditional identification methods based on various scales are not objective and accurate enough. Electroencephalogram (EEG) contains abundant physiological information, which makes it a new research direction to identify depression state. However, most EEG-based algorithms only extract the original EEG features and ignore the complex spatiotemporal information interactions, which will reduce performance. Thus, a more accurate and objective method for depression identification is urgently needed. In this work, we propose a novel depression identification model: W-GCN-GRU. In our proposed method, we censored six sensitive features based on Spearman's rank correlation coefficient and assigned different weight coefficients to each sensitive feature by AUC for the weighted fusion of sensitive features. In particular, we use the GCN and GRU cascade networks based on weighted sensitive features as depression recognition models. For the GCN, we creatively took the brain function network based on the correlation coefficient matrix as the adjacency matrix input and the weighted fused sensitive features were used as the node feature matrix input. Our proposed model performed well on our self-collected dataset and the MODMA datasets with a accuracy of 94.72%, outperforming other methods. Our findings showed that feature dimensionality reduction, weighted fusion, and EEG spatial information all had great effects on depression recognition.

Functional microRNA-targeting drug discovery by graph-based deep learning

MicroRNAs are recognized as key drivers in many cancers but targeting them with small molecules remains a challenge. We present RiboStrike, a deep-learning framework that identifies small molecules against specific microRNAs. To demonstrate its capabilities, we applied it to microRNA-21 (miR-21), a known driver of breast cancer. To ensure selectivity toward miR-21, we performed counter-screens against miR-122 and DICER. Auxiliary models were used to evaluate toxicity and rank the candidates. Learning from various datasets, we screened a pool of nine million molecules and identified eight, three of which showed anti-miR-21 activity in both reporter assays and RNA sequencing experiments. Target selectivity of these compounds was assessed using microRNA profiling and RNA sequencing analysis. The top candidate was tested in a xenograft mouse model of breast cancer metastasis, demonstrating a significant reduction in lung metastases. These results demonstrate RiboStrike's ability to nominate compounds that target the activity of miRNAs in cancer.

A Fault Diagnosis Method for 5G Cellular Networks Based on Knowledge and Data Fusion

As 5G networks become more complex and heterogeneous, the difficulty of network operation and maintenance forces mobile operators to find new strategies to stay competitive. However, most existing network fault diagnosis methods rely on manual testing and time stacking, which suffer from long optimization cycles and high resource consumption. Therefore, we herein propose a knowledge- and data-fusion-based fault diagnosis algorithm for 5G cellular networks from the perspective of big data and artificial intelligence. The algorithm uses a generative adversarial network (GAN) to expand the data set collected from real network scenarios to balance the number of samples under different network fault categories. In the process of fault diagnosis, a naive Bayesian model (NBM) combined with domain expert knowledge is firstly used to pre-diagnose the expanded data set and generate a topological association graph between the data with solid engineering significance and interpretability. Then, as the pre-diagnostic prior knowledge, the topological association graph is fed into the graph convolutional neural network (GCN) model simultaneously with the training data set for model training. We use a data set collected by Minimization of Drive Tests under real network scenarios in Lu'an City, Anhui Province, in August 2019. The simulation results demonstrate that the algorithm outperforms other traditional models in fault detection and diagnosis tasks, achieving an accuracy of 90.56% and a macro F1 score of 88.41%.

Parsing Netlists of Integrated Circuits from Images via Graph Attention Network

A massive number of paper documents that include important information such as circuit schematics can be converted into digital documents by optical sensors like scanners or digital cameras. However, extracting the netlists of analog circuits from digital documents is an exceptionally challenging task. This process aids enterprises in digitizing paper-based circuit diagrams, enabling the reuse of analog circuit designs and the automatic generation of datasets required for intelligent design models in this domain. This paper introduces a bottom-up graph encoding model aimed at automatically parsing the circuit topology of analog integrated circuits from images. The model comprises an improved electronic component detection network based on the Swin Transformer, an algorithm for component port localization, and a graph encoding model. The objective of the detection network is to accurately identify component positions and types, followed by automatic dataset generation through port localization, and finally, utilizing the graph encoding model to predict potential connections between circuit components. To validate the model's performance, we annotated an electronic component detection dataset and a circuit diagram dataset, comprising 1200 and 3552 training samples, respectively. Detailed experimentation results demonstrate the superiority of our proposed enhanced algorithm over comparative algorithms across custom and public datasets. Furthermore, our proposed port localization algorithm significantly accelerates the annotation speed of circuit diagram datasets.

Complexity-based graph convolutional neural network for epilepsy diagnosis in normal, acute, and chronic stages

Introduction

The automatic precision detection technology based on electroencephalography (EEG) is essential in epilepsy studies. It can provide objective proof for epilepsy diagnosis, treatment, and evaluation, thus helping doctors improve treatment efficiency. At present, the normal and acute phases of epilepsy can be well identified through EEG analysis, but distinguishing between the normal and chronic phases is still tricky.

Methods

In this paper, five popular complexity indicators of EEG signal, including approximate entropy, sample entropy, permutation entropy, fuzzy entropy and Kolmogorov complexity, are computed from rat hippocampi to characterize the normal, acute, and chronic phases during epileptogenesis. Results of one-way ANOVA and principal component analysis both show that utilizing complexity features, we are able to easily identify differences between normal, acute, and chronic phases. We also propose an innovative framework for epilepsy detection based on graph convolutional neural network (GCNN) using multi-channel EEG complexity as input.

Results

Combining information of five complexity measures at eight channels, our GCNN model demonstrate superior ability in recognizing the normal, acute, and chronic phases. Experiments results show that our GCNN model reached the high prediction accuracy above 98% and F1 score above 97% among these three phases for each individual rat.

Discussion

Our research practice based on real data shows that EEG complexity characteristics are of great significance for recognizing different stages of epilepsy.

DAHNGC: A Graph Convolution Model for Drug-Disease Association Prediction by Using Heterogeneous Network

In the field of drug development and repositioning, the prediction of drug-disease associations is a critical task. A recently proposed method for predicting drug-disease associations based on graph convolution relies heavily on the features of adjacent nodes within the homogeneous network for characterizing information. However, this method lacks node attribute information from heterogeneous networks, which could hardly provide valuable insights for predicting drug-disease associations. In this study, a novel drug-disease association prediction model called DAHNGC is proposed, which is based on a graph convolutional neural network. This model includes two feature extraction methods that are specifically designed to extract the attribute characteristics of drugs and diseases from both homogeneous and heterogeneous networks. First, the DropEdge technique is added to the graph convolutional neural network to alleviate the oversmoothing problem and obtain the characteristics of the same nodes of drugs or diseases in the homogeneous network. Then, an automatic feature extraction method in the heterogeneous network is designed to obtain the features of drugs or diseases at different nodes. Finally, the obtained features are put into the fully connected network for nonlinear transformation, and the potential drug-disease pairs are obtained by bilinear decoding. Experimental results demonstrate that the DAHNGC model exhibits good predictive performance for drug-disease associations.

Cellular Network Fault Diagnosis Method Based on a Graph Convolutional Neural Network

The efficient and accurate diagnosis of faults in cellular networks is crucial for ensuring smooth and uninterrupted communication services. In this paper, we propose an improved 4G/5G network fault diagnosis with a few effective labeled samples. Our solution is a heterogeneous wireless network fault diagnosis algorithm based on Graph Convolutional Neural Network (GCN). First, the common failure types of 4G/5G networks are analyzed, and then the graph structure is constructed with the data in the network parameter, given data sets as nodes and similarities as edges. GCN is used to extract features from the graph data, complete the classification task for nodes, and finally predict the fault types of cells. A large number of experiments are carried out based on the real data set, which is achieved by driving tests. The results show that, compared with a variety of traditional algorithms, the proposed method can effectively improve the performance of network fault diagnosis with a small number of labeled samples.

WPD-Enhanced Deep Graph Contrastive Learning Data Fusion for Fault Diagnosis of Rolling Bearing

Rolling bearings are crucial mechanical components in the mechanical industry. Timely intervention and diagnosis of system faults are essential for reducing economic losses and ensuring product productivity. To further enhance the exploration of unlabeled time-series data and conduct a more comprehensive analysis of rolling bearing fault information, this paper proposes a fault diagnosis technique for rolling bearings based on graph node-level fault information extracted from 1D vibration signals. In this technique, 10 categories of 1D vibration signals from rolling bearings are sampled using a sliding window approach. The sampled data is then subjected to wavelet packet decomposition (WPD), and the wavelet energy from the final layer of the four-level WPD decomposition in each frequency band is used as the node feature. The weights of edges between nodes are calculated using the Pearson correlation coefficient (PCC) to construct a node graph that describes the feature information of rolling bearings under different health conditions. Data augmentation of the node graph in the dataset is performed by randomly adding nodes and edges. The graph convolutional neural network (GCN) is employed to encode the augmented node graph representation, and deep graph contrastive learning (DGCL) is utilized for the pre-training and classification of the node graph. Experimental results demonstrate that this method outperforms contrastive learning-based fault diagnosis methods for rolling bearings and enables rapid fault diagnosis, thus ensuring the normal operation of mechanical systems. The proposed WPDPCC-DGCL method offers two advantages: (1) the flexibility of wavelet packet decomposition in handling non-smooth vibration signals and combining it with the powerful multi-scale feature encoding capability of GCN for richer characterization of fault information, and (2) the construction of graph node-level fault samples to effectively capture underlying fault information. The experimental results demonstrate the superiority of this method in rolling bearing fault diagnosis over contrastive learning-based approaches, enabling fast and accurate fault diagnoses for rolling bearings and ensuring the normal operation of mechanical systems.

Deciphering tissue heterogeneity from spatially resolved transcriptomics by the autoencoder-assisted graph convolutional neural network

Spatially resolved transcriptomics (SRT) provides an unprecedented opportunity to investigate the complex and heterogeneous tissue organization. However, it is challenging for a single model to learn an effective representation within and across spatial contexts. To solve the issue, we develop a novel ensemble model, AE-GCN (autoencoder-assisted graph convolutional neural network), which combines the autoencoder (AE) and graph convolutional neural network (GCN), to identify accurate and fine-grained spatial domains. AE-GCN transfers the AE-specific representations to the corresponding GCN-specific layers and unifies these two types of deep neural networks for spatial clustering via the clustering-aware contrastive mechanism. In this way, AE-GCN accommodates the strengths of both AE and GCN for learning an effective representation. We validate the effectiveness of AE-GCN on spatial domain identification and data denoising using multiple SRT datasets generated from ST, 10x Visium, and Slide-seqV2 platforms. Particularly, in cancer datasets, AE-GCN identifies disease-related spatial domains, which reveal more heterogeneity than histological annotations, and facilitates the discovery of novel differentially expressed genes of high prognostic relevance. These results demonstrate the capacity of AE-GCN to unveil complex spatial patterns from SRT data.

SGPPI: structure-aware prediction of protein-protein interactions in rigorous conditions with graph convolutional network

While deep learning (DL)-based models have emerged as powerful approaches to predict protein-protein interactions (PPIs), the reliance on explicit similarity measures (e.g. sequence similarity and network neighborhood) to known interacting proteins makes these methods ineffective in dealing with novel proteins. The advent of AlphaFold2 presents a significant opportunity and also a challenge to predict PPIs in a straightforward way based on monomer structures while controlling bias from protein sequences. In this work, we established Structure and Graph-based Predictions of Protein Interactions (SGPPI), a structure-based DL framework for predicting PPIs, using the graph convolutional network. In particular, SGPPI focused on protein patches on the protein-protein binding interfaces and extracted the structural, geometric and evolutionary features from the residue contact map to predict PPIs. We demonstrated that our model outperforms traditional machine learning methods and state-of-the-art DL-based methods using non-representation-bias benchmark datasets. Moreover, our model trained on human dataset can be reliably transferred to predict yeast PPIs, indicating that SGPPI can capture converging structural features of protein interactions across various species. The implementation of SGPPI is available at https://github.com/emerson106/SGPPI.

A New Regression Model for Depression Severity Prediction Based on Correlation among Audio Features Using a Graph Convolutional Neural Network

Recent studies have revealed mutually correlated audio features in the voices of depressed patients. Thus, the voices of these patients can be characterized based on the combinatorial relationships among the audio features. To date, many deep learning-based methods have been proposed to predict the depression severity using audio data. However, existing methods have assumed that the individual audio features are independent. Hence, in this paper, we propose a new deep learning-based regression model that allows for the prediction of depression severity on the basis of the correlation among audio features. The proposed model was developed using a graph convolutional neural network. This model trains the voice characteristics using graph-structured data generated to express the correlation among audio features. We conducted prediction experiments on depression severity using the DAIC-WOZ dataset employed in several previous studies. The experimental results showed that the proposed model achieved a root mean square error (RMSE) of 2.15, a mean absolute error (MAE) of 1.25, and a symmetric mean absolute percentage error of 50.96%. Notably, RMSE and MAE significantly outperformed the existing state-of-the-art prediction methods. From these results, we conclude that the proposed model can be a promising tool for depression diagnosis.

MiRNA-gene network embedding for predicting cancer driver genes

The development and progression of cancer arise due to the accumulation of mutations in driver genes. Correctly identifying the driver genes that lead to cancer development can significantly assist the drug design, cancer diagnosis and treatment. Most computer methods detect cancer drivers based on gene-gene networks by assuming that driver genes tend to work together, form protein complexes and enrich pathways. However, they ignore that microribonucleic acid (RNAs; miRNAs) regulate the expressions of their targeted genes and are related to human diseases. In this work, we propose a graph convolution network (GCN) approach called GM-GCN to identify the cancer driver genes based on a gene-miRNA network. First, we constructed a gene-miRNA network, where the nodes are miRNAs and their targeted genes. The edges connecting miRNA and genes indicate the regulatory relationship between miRNAs and genes. We prepared initial attributes for miRNA and genes according to their biological properties and used a GCN model to learn the gene feature representations in the network by aggregating the features of their neighboring miRNA nodes. And then, the learned features were passed through a 1D convolution module for feature dimensionality change. We employed the learned and original gene features to optimize model parameters. Finally, the gene features learned from the network and the initial input gene features were fed into a logistic regression model to predict whether a gene is a driver gene. We applied our model and state-of-the-art methods to predict cancer drivers for pan-cancer and individual cancer types. Experimental results show that our model performs well in terms of the area under the receiver operating characteristic curve and the area under the precision-recall curve compared to state-of-the-art methods that work on gene networks. The GM-GCN is freely available via https://github.com/weiba/GM-GCN.

Classification of Depression and Its Severity Based on Multiple Audio Features Using a Graphical Convolutional Neural Network

Audio features are physical features that reflect single or complex coordinated movements in the vocal organs. Hence, in speech-based automatic depression classification, it is critical to consider the relationship among audio features. Here, we propose a deep learning-based classification model for discriminating depression and its severity using correlation among audio features. This model represents the correlation between audio features as graph structures and learns speech characteristics using a graph convolutional neural network. We conducted classification experiments in which the same subjects were allowed to be included in both the training and test data (Setting 1) and the subjects in the training and test data were completely separated (Setting 2). The results showed that the classification accuracy in Setting 1 significantly outperformed existing state-of-the-art methods, whereas that in Setting 2, which has not been presented in existing studies, was much lower than in Setting 1. We conclude that the proposed model is an effective tool for discriminating recurring patients and their severities, but it is difficult to detect new depressed patients. For practical application of the model, depression-specific speech regions appearing locally rather than the entire speech of depressed patients should be detected and assigned the appropriate class labels.

Granger-Causality-Based Multi-Frequency Band EEG Graph Feature Extraction and Fusion for Emotion Recognition

Graph convolutional neural networks (GCN) have attracted much attention in the task of electroencephalogram (EEG) emotion recognition. However, most features of current GCNs do not take full advantage of the causal connection between the EEG signals in different frequency bands during the process of constructing the adjacency matrix. Based on the causal connectivity between the EEG channels obtained by Granger causality (GC) analysis, this paper proposes a multi-frequency band EEG graph feature extraction and fusion method for EEG emotion recognition. First, the original GC matrices between the EEG signals at each frequency band are calculated via GC analysis, and then they are adaptively converted to asymmetric binary GC matrices through an optimal threshold. Then, a kind of novel GC-based GCN feature (GC-GCN) is constructed by using differential entropy features and the binary GC matrices as the node values and adjacency matrices, respectively. Finally, on the basis of the GC-GCN features, a new multi-frequency band feature fusion method (GC-F-GCN) is proposed, which integrates the graph information of the EEG signals at different frequency bands for the same node. The experimental results demonstrate that the proposed GC-F-GCN method achieves better recognition performance than the state-of-the-art GCN methods, for which average accuracies of 97.91%, 98.46%, and 98.15% were achieved for the arousal, valence, and arousal-valence classifications, respectively.

Research of Software Defect Prediction Model Based on Complex Network and Graph Neural Network

The goal of software defect prediction is to make predictions by mining the historical data using models. Current software defect prediction models mainly focus on the code features of software modules. However, they ignore the connection between software modules. This paper proposed a software defect prediction framework based on graph neural network from a complex network perspective. Firstly, we consider the software as a graph, where nodes represent the classes, and edges represent the dependencies between the classes. Then, we divide the graph into multiple subgraphs using the community detection algorithm. Thirdly, the representation vectors of the nodes are learned through the improved graph neural network model. Lastly, we use the representation vector of node to classify the software defects. The proposed model is tested on the PROMISE dataset, using two graph convolution methods, based on the spectral domain and spatial domain in the graph neural network. The investigation indicated that both convolution methods showed an improvement in various metrics, such as accuracy, F-measure, and MCC (Matthews correlation coefficient) by 86.6%, 85.8%, and 73.5%, and 87.5%, 85.9%, and 75.5%, respectively. The average improvement of various metrics was noted as 9.0%, 10.5%, and 17.5%, and 6.3%, 7.0%, and 12.1%, respectively, compared with the benchmark models.

SCDD: a novel single-cell RNA-seq imputation method with diffusion and denoising

Single-cell sequencing technologies are widely used to discover the evolutionary relationships and the differences in cells. Since dropout events may frustrate the analysis, many imputation approaches for single-cell RNA-seq data have appeared in previous attempts. However, previous imputation attempts usually suffer from the over-smooth problem, which may bring limited improvement or negative effect for the downstream analysis of single-cell RNA-seq data. To solve this difficulty, we propose a novel two-stage diffusion-denoising method called SCDD for large-scale single-cell RNA-seq imputation in this paper. We introduce the diffusion i.e. a direct imputation strategy using the expression of similar cells for potential dropout sites, to perform the initial imputation at first. After the diffusion, a joint model integrated with graph convolutional neural network and contractive autoencoder is developed to generate superposition states of similar cells, from which we restore the original states and remove the noise introduced by the diffusion. The final experimental results indicate that SCDD could effectively suppress the over-smooth problem and remarkably improve the effect of single-cell RNA-seq downstream analysis, including clustering and trajectory analysis.

Multi-Geometric Reasoning Network for Insulator Defect Detection of Electric Transmission Lines

To address the challenges in the unmanned system-based intelligent inspection of electric transmission line insulators, this paper proposed a multi-geometric reasoning network (MGRN) to accurately detect insulator geometric defects based on aerial images with complex backgrounds and different scales. The spatial geometric reasoning sub-module (SGR) was developed to represent the spatial location relationship of defects. The appearance geometric reasoning sub-module (AGR) and the parallel feature transformation (PFT) sub-module were adopted to obtain the appearance geometric features from the real samples. These multi-geometric features can be fused with the original visual features to identify and locate the insulator defects. The proposed solution is assessed through experiments against the existing solutions and the numerical results indicate that it can significantly improve the detection accuracy of multiple insulator defects using the aerial images.

Spatial-temporal graph convolutional network for Alzheimer classification based on brain functional connectivity imaging of electroencephalogram

Functional connectivity of the human brain, representing statistical dependence of information flow between cortical regions, significantly contributes to the study of the intrinsic brain network and its functional mechanism. To fully explore its potential in the early diagnosis of Alzheimer's disease (AD) using electroencephalogram (EEG) recordings, this article introduces a novel dynamical spatial-temporal graph convolutional neural network (ST-GCN) for better classification performance. Different from existing studies that are based on either topological brain function characteristics or temporal features of EEG, the proposed ST-GCN considers both the adjacency matrix of functional connectivity from multiple EEG channels and corresponding dynamics of signal EEG channel simultaneously. Different from the traditional graph convolutional neural networks, the proposed ST-GCN makes full use of the constrained spatial topology of functional connectivity and the discriminative dynamic temporal information represented by the 1D convolution. We conducted extensive experiments on the clinical EEG data set of AD patients and Healthy Controls. The results demonstrate that the proposed method achieves better classification performance (92.3%) than the state-of-the-art methods. This approach can not only help diagnose AD but also better understand the effect of normal ageing on brain network characteristics before we can accurately diagnose the condition based on resting-state EEG.

Predicting CT-Based Coronary Artery Disease Using Vascular Biomarkers Derived from Fundus Photographs with a Graph Convolutional Neural Network

The study population contains 145 patients who were prospectively recruited for coronary CT angiography (CCTA) and fundoscopy. This study first examined the association between retinal vascular changes and the Coronary Artery Disease Reporting and Data System (CAD-RADS) as assessed on CCTA. Then, we developed a graph neural network (GNN) model for predicting the CAD-RADS as a proxy for coronary artery disease. The CCTA scans were stratified by CAD-RADS scores by expert readers, and the vascular biomarkers were extracted from their fundus images. Association analyses of CAD-RADS scores were performed with patient characteristics, retinal diseases, and quantitative vascular biomarkers. Finally, a GNN model was constructed for the task of predicting the CAD-RADS score compared to traditional machine learning (ML) models. The experimental results showed that a few retinal vascular biomarkers were significantly associated with adverse CAD-RADS scores, which were mainly pertaining to arterial width, arterial angle, venous angle, and fractal dimensions. Additionally, the GNN model achieved a sensitivity, specificity, accuracy and area under the curve of 0.711, 0.697, 0.704 and 0.739, respectively. This performance outperformed the same evaluation metrics obtained from the traditional ML models (p < 0.05). The data suggested that retinal vasculature could be a potential biomarker for atherosclerosis in the coronary artery and that the GNN model could be utilized for accurate prediction.

Study of crystal properties based on attention mechanism and crystal graph convolutional neural network

The prediction of crystal properties has always been limited by huge computational costs. In recent years, the rise of machine learning methods has gradually made it possible to study crystal properties on a large scale. We propose an attention mechanism-based crystal graph convolutional neural network, which builds a machine learning model by inputting crystallographic information files and target properties. In our research, the attention mechanism is introduced in the crystal graph convolutional neural network (CGCNN) to learn the local chemical environment, and node normalization is added to reduce the risk of overfitting. We collect structural information and calculation data of about 36 000 crystals and examine the prediction performance of the models for the formation energy, total energy, bandgap, and Fermi energy of crystals in our research. Compared with the CGCNN, it is found that the accuracy (ACCU) of the predicted properties can be further improved to varying degrees by the introduction of the attention mechanism. Moreover, the total magnetization and bandgap can be classified under the same neural network framework. The classification ACCU of wide bandgap semiconductor crystals with a bandgap threshold of 2.3 eV reaches 93.2%, and the classification ACCU of crystals with a total magnetization threshold of 0.5 μ_Breaches 88.8%. The work is helpful to realize large-scale prediction and classification of crystal properties, accelerating the discovery of new functional crystal materials.

Predicting Brain Amyloid-β PET Grades with Graph Convolutional Networks Based on Functional MRI and Multi-Level Functional Connectivity

BACKGROUND

The detection of amyloid-β (Aβ) deposition in the brain provides crucial evidence in the clinical diagnosis of Alzheimer's disease (AD). However, the current positron emission tomography (PET)-based brain Aβ examination suffers from the problems of coarse visual inspection (in many cases, with 2-class stratification) and high scanning cost.

OBJECTIVE

1) To characterize the non-binary Aβ deposition levels in the AD continuum based on clustering of PET data, and 2) to explore the feasibility of predicting individual Aβ deposition grades with non-invasive functional magnetic resonance imaging (fMRI).

METHODS

1) Individual whole-brain Aβ-PET images from the OASIS-3 dataset (N = 258) were grouped into three clusters (grades) with t-SNE and k-means. The demographical data as well as global and regional standard uptake value ratios (SUVRs) were compared among the three clusters with Chi-square tests or ANOVA tests. 2) From resting-state fMRI, both conventional functional connectivity (FC) and high-order FC networks were constructed and the topological architectures of the two networks were jointly learned with graph convolutional networks (GCNs) to predict the Aβ-PET grades for each individual.

RESULTS

We found three clearly separated clusters, indicating three Aβ-PET grades. There were significant differences in gender, age, cognitive ability, APOE type, as well as global and regional SUVRs among the three grades we found. The prediction of Aβ-PET grades with GCNs on FC for the 258 participants in the AD continuum reached a satisfactory averaged accuracy (78.8%) in the two-class classification tasks.

CONCLUSION

The results demonstrated the feasibility of using deep learning on a non-invasive brain functional imaging technique to approximate PET-based Aβ deposition grading.

A Plant Disease Recognition Method Based on Fusion of Images and Graph Structure Text

The disease image recognition models based on deep learning have achieved relative success under limited and restricted conditions, but such models are generally subjected to the shortcoming of weak robustness. The model accuracy would decrease obviously when recognizing disease images with complex backgrounds under field conditions. Moreover, most of the models based on deep learning only involve characterization learning on visual information in the image form, while the expression of other modal information rather than the image form is often ignored. The present study targeted the main invasive diseases in tomato and cucumber as the research object. Firstly, in response to the problem of weak robustness, a feature decomposition and recombination method was proposed to allow the model to learn image features at different granularities so as to accurately recognize different test images. Secondly, by extracting the disease feature words from the disease text description information composed of continuous vectors and recombining them into the disease graph structure text, the graph convolutional neural network (GCN) was then applied for feature learning. Finally, a vegetable disease recognition model based on the fusion of images and graph structure text was constructed. The results show that the recognition accuracy, precision, sensitivity, and specificity of the proposed model were 97.62, 92.81, 98.54, and 93.57%, respectively. This study improved the model robustness to a certain extent, and provides ideas and references for the research on the fusion method of image information and graph structure information in disease recognition.

Improving cancer driver gene identification using multi-task learning on graph convolutional network

Cancer is thought to be caused by the accumulation of driver genetic mutations. Therefore, identifying cancer driver genes plays a crucial role in understanding the molecular mechanism of cancer and developing precision therapies and biomarkers. In this work, we propose a Multi-Task learning method, called MTGCN, based on the Graph Convolutional Network to identify cancer driver genes. First, we augment gene features by introducing their features on the protein-protein interaction (PPI) network. After that, the multi-task learning framework propagates and aggregates nodes and graph features from input to next layer to learn node embedding features, simultaneously optimizing the node prediction task and the link prediction task. Finally, we use a Bayesian task weight learner to balance the two tasks automatically. The outputs of MTGCN assign each gene a probability of being a cancer driver gene. Our method and the other four existing methods are applied to predict cancer drivers for pan-cancer and some single cancer types. The experimental results show that our model shows outstanding performance compared with the state-of-the-art methods in terms of the area under the Receiver Operating Characteristic (ROC) curves and the area under the precision-recall curves. The MTGCN is freely available via https://github.com/weiba/MTGCN.

ELASPIC2 (EL2): Combining Contextualized Language Models and Graph Neural Networks to Predict Effects of Mutations

The ELASPIC web server allows users to evaluate the effect of mutations on protein folding and protein-protein interaction on a proteome-wide scale. It uses homology models of proteins and protein-protein interactions, which have been precalculated for several proteomes, and machine learning models, which integrate structural information with sequence conservation scores, in order to make its predictions. Since the original publication of the ELASPIC web server, several advances have motivated a revisiting of the problem of mutation effect prediction. First, progress in neural network architectures and self-supervised pre-trained has resulted in models which provide more informative embeddings of protein sequence and structure than those used by the original version of ELASPIC. Second, the amount of training data has increased several-fold, largely driven by advances in deep mutation scanning and other multiplexed assays of variant effect. Here, we describe two machine learning models which leverage the recent advances in order to achieve superior accuracy in predicting the effect of mutation on protein folding and protein-protein interaction. The models incorporate features generated using pre-trained transformer- and graph convolution-based neural networks, and are trained to optimize a ranking objective function, which permits the use of heterogeneous training data. The outputs from the new models have been incorporated into the ELASPIC web server, available at http://elaspic.kimlab.org.

Predicting the Associations between Meridians and Chinese Traditional Medicine Using a Cost-Sensitive Graph Convolutional Neural Network

Natural products are the most important and commonly used in Traditional Chinese Medicine (TCM) for healthcare and disease prevention in East-Asia. Although the Meridian system of TCM was established several thousand years ago, the rationale of Meridian classification based on the ingredient compounds remains poorly understood. A core challenge for the traditional machine learning approaches for chemical activity prediction is to encode molecules into fixed length vectors but ignore the structural information of the chemical compound. Therefore, we apply a cost-sensitive graph convolutional neural network model to learn local and global topological features of chemical compounds, and discover the associations between TCM and their Meridians. In the experiments, we find that the performance of our approach with the area under the receiver operating characteristic curve (ROC-AUC) of 0.82 which is better than the traditional machine learning algorithm and also obtains 8%–13% improvement comparing with the state-of-the-art methods. We investigate the powerful ability of deep learning approach to learn the proper molecular descriptors for Meridian prediction and to provide novel insights into the complementary and alternative medicine of TCM.

Prediction of the Health Effects of Food Peptides and Elucidation of the Mode-of-action Using Multi-task Graph Convolutional Neural Network

Food proteins work not only as nutrients but also modulators for the physiological functions of the human body. The physiological functions of food proteins are basically regulated by peptides encrypted in food protein sequences (food peptides). In this study, we propose a novel deep learning-based method to predict the health effects of food peptides and elucidate the mode-of-action. In the algorithm, we estimate potential target proteins of food peptides using a multi-task graph convolutional neural network, and predict its health effects using information about therapeutic targets for diseases. We constructed predictive models based on 21,103 peptide-protein interactions involving 10,950 peptides and 2,533 proteins, and applied the models to food peptides (e. g., lactotripeptide, isoleucyltyrosine and sardine peptide) defined in food for specified health use. The models suggested potential effects such as blood-pressure lowering effects, blood glucose level lowering effects, and anti-cancer effects for several food peptides. The interactions of food peptides with target proteins were confirmed by docking simulations.