Collaborative Decision-Reinforced Self-Supervision for Attributed Graph Clustering.

Deep node clustering based on mutual information maximization

In the ever-evolving landscape of data analysis, the need to efficiently and accurately interpret multimodal time series data has become paramount. Traditional methods often fall short in addressing the complex dependencies and dynamics inherent in such data, limiting their effectiveness in real-world applications. This work introduces a comprehensive approach that leverages Graph Attention Networks (GATs), Variational Graph Autoencoders (VGAEs), transfer learning with pretrained transformers, and Bayesian state-space models to overcome these limitations. GATs are selected for their ability to dynamically focus on relevant modalities through attention mechanisms, thereby capturing the intricate relationships between different data modalities. This method significantly enhances the model's ability to integrate multimodal information, leading to notable improvements in classification, prediction, and anomaly detection tasks. VGAEs are utilized to learn latent representations within a graph-based framework, promoting unsupervised learning while unveiling the underlying data structure. The resultant embeddings are pivotal for downstream tasks like clustering and visualization, encapsulating the interactions within multimodal time series data effectively. Furthermore, this work incorporates transfer learning with pretrained transformers to harness extensive knowledge from large datasets, adapting it to multimodal time series analysis. This strategy excels in capturing long-range dependencies, thereby augmenting generalization and performance in data-scarce scenarios. Bayesian state-space models are employed to elucidate the temporal dynamics and uncertainties of time series data, offering a robust framework for probabilistic inference and enhancing the interpretability and reliability of forecasting and anomaly detection. The efficacy of the proposed model is rigorously evaluated using diverse datasets, including the Yahoo! Stock Dataset, Forest Cover Dataset, and an empirical collection of 100k time series data samples. The results demonstrate a significant leap in performance metrics, including a 9.5% increase in precision, 8.5% boost in accuracy, 8.3% rise in recall, 10.4% reduction in delay, 9.4% enhancement in AUC, and a 5.9% improvement in specificity, alongside superior pre-emption capabilities compared to existing methods. This work not only addresses the pressing need for advanced multimodal time series analysis techniques but also sets a new benchmark for efficiency and accuracy. The integration of GATs, VGAEs, transfer learning with pretrained transformers, and Bayesian state-space models presents a formidable approach that significantly advances the field, offering profound impacts on a wide array of applications.

Investigating potential interactions between drugs and herbal medicines helps optimize combined treatment strategies and supports personalized and precision medicine. Deep learning-based methods have been successful in predicting drug-related interactions. However, these methods face challenges such as low data quality and uneven distribution. Large language models (LLMs) effectively address these challenges through their extensive knowledge bases. Motivated by this, we integrate LLMs, one-hot encoding, and variational graph autoencoders (VGAEs) to propose a herbal medicine-drug interaction (HDI) prediction model. First, LLMs are employed to extract features from drug SMILES, generating high-quality molecular representations. Second, one-hot encoding is applied to herbal medicines with multiple natural products to construct feature vectors and improve model interpretability. Finally, VGAEs are utilized to reconstruct herbal medicine-drug graphs and predict unknown HDIs. Additionally, we differentiate between herbal medicine-drug similarity and the degree of individual drug or herbal medicine nodes to mitigate the dominance of high-degree nodes in VGAE message flow. Multiple experiments were conducted to validate the significance of the proposed model and its key components. This method shows great potential for applications in traditional Chinese medicine formulation optimization, new drug development, and precision medicine.

BackgroundEnormous clinical and biomedical researches have demonstrated that microbes are crucial to human health. Identifying associations between microbes and diseases can not only reveal potential disease mechanisms, but also facilitate early diagnosis and promote precision medicine. Due to the data perturbation and unsatisfactory latent representation, there is a significant room for improvement.ResultsIn this work, we proposed a novel framework, Multi-scale Variational Graph AutoEncoder embedding Wasserstein distance (MVGAEW) to predict disease-related microbes, which had the ability to resist data perturbation and effectively generate latent representations for both microbes and diseases from the perspective of distribution. First, we calculated multiple similarities and integrated them through similarity network confusion. Subsequently, we obtained node latent representations by improved variational graph autoencoder. Ultimately, XGBoost classifier was employed to predict potential disease-related microbes. We also introduced multi-order node embedding reconstruction to enhance the representation capacity. We also performed ablation studies to evaluate the contribution of each section of our model. Moreover, we conducted experiments on common drugs and case studies, including Alzheimer’s disease, Crohn’s disease, and colorectal neoplasms, to validate the effectiveness of our framework.ConclusionsSignificantly, our model exceeded other currently state-of-the-art methods, exhibiting a great improvement on the HMDAD database.

In this paper, we introduce Gene Knockout Inference (GenKI), a virtual knockout (KO) tool for gene function prediction using single-cell RNA sequencing (scRNA-seq) data in the absence of KO samples when only wild-type (WT) samples are available. Without using any information from real KO samples, GenKI is designed to capture shifting patterns in gene regulation caused by the KO perturbation in an unsupervised manner and provide a robust and scalable framework for gene function studies. To achieve this goal, GenKI adapts a variational graph autoencoder (VGAE) model to learn latent representations of genes and interactions between genes from the input WT scRNA-seq data and a derived single-cell gene regulatory network (scGRN). The virtual KO data is then generated by computationally removing all edges of the KO gene—the gene to be knocked out for functional study—from the scGRN. The differences between WT and virtual KO data are discerned by using their corresponding latent parameters derived from the trained VGAE model. Our simulations show that GenKI accurately approximates the perturbation profiles upon gene KO and outperforms the state-of-the-art under a series of evaluation conditions. Using publicly available scRNA-seq data sets, we demonstrate that GenKI recapitulates discoveries of real-animal KO experiments and accurately predicts cell type-specific functions of KO genes. Thus, GenKI provides an in-silico alternative to KO experiments that may partially replace the need for genetically modified animals or other genetically perturbed systems.

BackgroundProtein-protein interactions (PPIs) play a critical role in essential biological processes such as signal transduction, enzyme activity regulation, cytoskeletal structure, immune responses, and gene regulation. However, current methods mainly focus on extracting features from protein sequences and using graph neural network (GNN) to acquire interaction information from the PPI network graph. This limits the model’s ability to learn richer and more effective interaction information, thereby affecting prediction performance.ResultsIn this study, we propose a novel deep learning method, MESM, for effectively predicting PPI. The datasets used for the PPI prediction task were primarily constructed from the STRING database, including two Homo sapiens PPI datasets, SHS27k and SHS148k, and two Saccharomyces cerevisiae PPI datasets, SYS30k and SYS60k. MESM consists of three key modules, as follows: First, MESM extracts multimodal representations from protein sequence information, protein structure information, and point cloud features through Sequence Variational Autoencoder (SVAE), Variational Graph Autoencoder (VGAE), and PointNet Autoencoder (PAE). Then, Fusion Autoencoder (FAE) is used to integrate these multimodal features, generating rich and balanced protein representations. Next, MESM leverages GraphGPS to learn structural information from the PPI network graph structure and combines Graph Attention Network (GAT) to further capture protein interaction information. Finally, MESM uses Graph Convolutional Network (GCN) and SubgraphGCN to extract global and local features from the perspective of the overall graph and subgraphs. Moreover, we build seven independent graphs from the overall PPI network graph to specifically learn the features of each PPI type, thereby enhancing the model’s learning ability for different types of interactions.ConclusionsCompared to the state-of-the-art methods, MESM achieved improvements of 8.77%, 4.98%, 7.48%, and 6.08% on SHS27k, SHS148k, SYS30k, and SYS60k, respectively. The experimental results demonstrate that MESM exhibits significant improvements in PPI prediction performance.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12915-025-02356-y.

Attributed graph clustering (AGC) is an important problem in graph mining as more and more complex data in real-world have been represented in graphs with attributed nodes. While it is a common practice to leverage both attribute and structure information for improved clustering performance, most existing AGC algorithms consider only a specific type of relations, which hinders their applicability to integrate various complex relations into node attributes for AGC. In this article, we propose GRACE, an extended graph convolution framework for AGC tasks. Our framework provides a general and interpretative solution for clustering many different types of attributed graphs, including undirected, directed, heterogeneous and hyper attributed graphs. By building suitable graph Laplacians for each of the aforementioned graph types, GRACE can seamlessly perform graph convolution on node attributes to fuse all available information for clustering. We conduct extensive experiments on 14 real-world datasets of four different graph types. The experimental results show that GRACE outperforms the state-of-the-art AGC methods on the different graph types in terms of clustering quality, time, and memory usage.

Graph convolutional autoencoders with co-learning of graph structure and node attributes

Graph representation learning has shown superior performance in numerous real-world applications, such as finance and social networks. Nevertheless, most existing works might make discriminatory predictions due to insufficient attention to fairness in their decision-making processes. This oversight has prompted a growing focus on fair representation learning. Among recent explorations on fair representation learning, prior works based on the adversarial learning usually induce unstable or counterproductive performance. To achieve fairness in a stable manner, we present the design and implementation of graph representation learning based on fairness information bottleneck (GRAFair), a new framework based on a variational graph autoencoder (VGAE). The crux of GRAFair is the conditional fairness bottleneck (CFB), where the objective is to capture the trade-off between the utility of representations and sensitive information of interest. By applying variational approximation, we can make the optimization objective tractable. Particularly, GRAFair can be trained to produce informative representations of tasks while containing little sensitive information without adversarial training. Experiments on various real-world datasets demonstrate the effectiveness of our proposed method in terms of fairness, utility, robustness, and stability.

The distribution network is the main component of the power system, which undertakes the important function of power transmission and distribution. Fast and accurate distribution network fault location is one of the important means to ensure the safe and stable operation of the power system and is helpful in guiding troubleshooting, shortening the power outage time, reducing the workload of manual inspection, and reducing the social and economic losses caused by faults. Due to the development of the new distribution network, the comprehensive influence of many factors has put forward new challenges to the traditional distribution network fault location methods. In this paper, a distribution network fault location method based on the graph neural network is proposed. Firstly, the distribution network is treated as non-Euclidean graph data; secondly, variational graph auto-encoders (VGAE) are used to mine the underlying information of nodes and improve the overall anti-noise performance of the fault location method. Then the GraphSAGE model is used to aggregate the neighbor information of nodes, fully consider the influence of the surrounding lines on the target lines, and improve the output of the model to locate the distribution network line where the fault occurred. The experimental example analysis based on OpenDSS simulation software (version 9.8) proves that the proposed method has high accuracy and anti-interference, and the accuracy reached 97.81%. Moreover, the positioning result is still good in the new intelligent distribution network scenario with distributed power access, with an accuracy of 95.07% in the hybrid power generation scenario.

Ethereum, the second-largest cryptocurrency after Bitcoin, has attracted wide attention in the last few years and accumulated significant transaction records. However, the underlying Ethereum network structure is still relatively unexplored. Also, very few attempts have been made to perform link predictability on the Ethereum transactions network. This paper presents a Detailed Analysis of the Ethereum Network on Transaction Behavior, Community Structure, and Link Prediction (DANET) framework to investigate various valuable aspects of the Ethereum network. Specifically, we explore the change in wealth distribution and accumulation on Ethereum Featured Transactional Network (EFTN) and further study its community structure. We further hunt for a suitable link predictability model on EFTN by employing state-of-the-art Variational Graph Auto-Encoders. The link prediction experimental results demonstrate the superiority of outstanding prediction accuracy on Ethereum networks. Moreover, the statistic usages of the Ethereum network are visualized and summarized through the experiments allowing us to formulate conjectures on the current use of this technology and future development.

Ethereum, the second-largest cryptocurrency after Bitcoin, has attracted wide attention in the last few years and accumulated significant transaction records. However, the underlying Ethereum network structure is still relatively unexplored. Also, very few attempts have been made to perform link predictability on the Ethereum transactions network. This paper presents a Detailed Analysis of the Ethereum Network on Transaction Behavior, Community Structure, and Link Prediction (DANET) framework to investigate various valuable aspects of the Ethereum network. Specifically, we explore the change in wealth distribution and accumulation on Ethereum Featured Transactional Network (EFTN) and further study its community structure. We further hunt for a suitable link predictability model on EFTN by employing state-of-the-art Variational Graph Auto-Encoders. The link prediction experimental results demonstrate the superiority of outstanding prediction accuracy on Ethereum networks. Moreover, the statistic usages of the Ethereum network are visualized and summarized through the experiments allowing us to formulate conjectures on the current use of this technology and future development.

Deciphering gene regulatory networks (GRNs) from single-cell transcriptomics data remains a fundamental challenge in computational biology. It is hindered by data sparsity, high dimensionality, and the lack of scalable, generalizable inference models. To address this, we present GRNFormer, a generalizable graph transformer framework for accurate GRN inference from transcriptomics data across species, cell types, and platforms without requiring cell-type annotations or prior regulatory information. GRNFormer integrates a transformer-based gene expression encoder (Gene-Transcoder) with a variational graph autoencoder (GraViTAE) employing pairwise attention to jointly learn the representations of genes (nodes) and their co-expression relationships (edges). Leveraging TF-Walker, a transcription factor-anchored subgraph sampling strategy, it effectively captures gene regulatory interactions from either single-cell or bulk RNA-seq datasets. Benchmarking on standard datasets demonstrates that GRNFormer outperforms existing traditional and deep learning state-of-the-art methods in blind evaluations, achieving average sampled area under the receiver operating characteristic curve (Sampled_AUROC) and sampled area under the precision-recall curve (Sampled_AUPRC) values between 0.90 and 0.98 as well as 0.87-0.98 average sampled F1 score. The model robustly recovers both known and novel regulatory networks, including pluripotency circuits in human embryonic stem cells (hESCs) and immune cell modules in peripheral blood mononuclear cells (PBMCs). The architecture enables scalable, biologically interpretable GRN inference across various datasets, cell types, and species, establishing GRNFormer as a robust and transferable tool for network biology. GRNFormer is available on GitHub (https://github.com/BioinfoMachineLearning/GRNformer); the version used in this work is archived on Zenodo (https://doi.org/10.5281/zenodo.18868395), with evaluation resources for reproducibility.

Graph-structured data contains a lot of sensitive information about individuals. In order to protect users' privacy, many anonymization mechanisms for graph-structured data are proposed. However, one common drawback of these mechanisms is that they only consider to hide the local characteristics, such as the degree of nodes or their neighbors. They lack the consideration for the nodes' attribute features and the features of potential global graph structure, which leads to the failure of these mechanisms to provide sufficient security.To address this shortcoming, we propose RecGraph, a framework for graph recovery attack based on variational graph autoencoders. We use RecGraph to perform graph recovery attack on three real social network datasets, and compare it with five existing baselines, to prove the effectiveness of our method. We also evaluate the privacy wastage after performing the graph recovery attack using RecGraph to demonstrate the serious security risks faced by the existing graph anonymization mechanisms.

In this paper, we propose and compare two novel deep generative model-based approaches for the design representation, reconstruction, and generation of porous metamaterials characterized by complex and fully connected solid and pore networks. A highly diverse porous metamaterial database is curated, with each sample represented by solid and pore phase graphs and a voxel image. All metamaterial samples adhere to the requirement of complete connectivity in both pore and solid phases. The first approach employs a Dual Decoder Variational Graph Autoencoder to generate both solid phase and pore phase graphs. The second approach employs a Variational Graph Autoencoder for reconstructing/generating the nodes in the solid phase and pore phase graphs and a Transformer-based Large Language Model (LLM) for reconstructing/generating the connections, i.e., the edges among the nodes. A comparative study was conducted, and we found that both approaches achieved high accuracy in reconstructing node features, while the LLM exhibited superior performance in reconstructing edge features. Reconstruction accuracy is also validated by voxel-to-voxel comparison between the reconstructions and the original images in the test set. Additionally, discussions on the advantages and limitations of using LLMs in metamaterial design generation, along with the rationale behind their utilization, are provided.