MOTIVATION: Spatial transcriptomics (ST) enables the study of spatial heterogeneity in tissues. However, current methods struggle with large-scale, high-resolution data, leading to reduced efficiency and accuracy in detecting spatial domains. A scalable, precise solution is urgently needed. RESULTS: We present STAHD, a scalable and efficient framework for spatial domain detection in ST data. Combining a graph attention autoencoder with multilevel k-way graph partitioning, STAHD decomposes large graphs into compact subgraphs and generates low-dimensional embeddings. This improves computational efficiency and clustering accuracy. Benchmarks on human and mouse datasets show STAHD outperforms existing methods and accurately reveals spatially distinct tumor microenvironments and functional regions. AVAILABILITY AND IMPLEMENTATION: Source code and data are available at: https://github.com/Little-Eel/STAHD.

The breakthrough development of single-cell RNA sequencing technology enables tissue heterogeneity analysis at single-cell resolution, where accurate cell type annotation is crucial for unlocking its full potential. To address three key challenges in current annotation methods-scarce labeled data, suboptimal graph topology, and missing cell state information-we propose scALGSL, an innovative framework integrating dynamic graph optimization with active learning. Our core contributions are threefold: (1) A graph-guided active learning mechanism adaptively selects high-value training samples, significantly alleviating label scarcity; (2) A learnable graph structure optimization module dynamically refines adjacency matrices to eliminate spurious connections caused by data sparsity; (3) A novel cell state auxiliary pathway extracts critical functional features via pre-trained models to enhance type discrimination. The systematic review showed that the average accuracy and f1 of scALGSL on the cancer dataset were 0.896 and 0.771, respectively, and it showed good robustness in cross-platform tasks. Integration of cell state information substantially boosts performance, while ablation studies validate the necessity of node selection and edge optimization modules. This framework provides a scalable solution for precise cell annotation, facilitating tumor microenvironment analysis and precision medicine applications.

Predicting unknown transcription factor-target gene (TF-target gene) interactions based on gene regulatory networks (GRNs) is critical for understanding cellular functions and biological discovery. Existing computational methods focus on the statistic corelation and neglect the semantic information of biomedical knowledge. To address this, we propose BioLinkGPT, a novel framework that integrates the robust textual comprehension capabilities of Large Language Models (LLMs) with graph neural networks (GNNs) to capture structural information of GRN. BioLinkGPT leverages gene information and the known TF-target gene relationships from literature to predict unknown TF-target gene interactions and improve its performance via a two-stage instruction fine-tuning strategy. We construct a comprehensive GRN dataset focused on human infectious diseases, and experimental results show that BioLinkGPT significantly outperforms existing baselines in link prediction metrics. We validate the model's accurate identification of known regulatory relationships and find that BioLinkGPT successfully uncover potential regulatory interactions with high confidence and biological relevance. BioLinkGPT offers an efficient and precise computational tool to infer TF-target gene interactions, accelerating research on infectious disease mechanisms and drug target development.

Phages, viruses that infect bacteria, offer a promising strategy against antibiotic-resistant pathogens. Phage viral proteins (PVPs) are essential for phage-host interactions, yet their identification and functional annotation remain challenging due to high sequence diversity, limited experimental data, and class imbalance. To address these issues, we propose ProtPhage, a novel framework that leverages the ProtT5 protein language model for richer sequence representation compared to traditional methods. Additionally, ProtPhage incorporates an asymmetric loss function to mitigate class imbalance, significantly improving the prediction of the minority class "minor capsid," with an F1 score 33.07$\%$ higher than the best existing model. Extensive experiments demonstrate that ProtPhage outperforms current methods across multiple metrics, including accuracy, precision, recall, and F1 score. A case study on the Mycobacterium phage PDRPxv genome further validates its practical utility, while expanded experiments highlight its potential in phage-host prediction. By integrating advanced deep learning techniques, ProtPhage establishes a new standard for PVP identification and annotation, contributing to the broader field of computational phage biology.

Drug-target interactions (DTIs) play a key role in drug development, and accurate prediction can significantly improve the efficiency of this process. Traditional experimental methods are reliable but time-consuming and laborious. With the rapid development of deep learning, many DTI prediction methods have emerged. However, most of these methods only focus on the intrinsic features of drugs and targets, while ignoring the relational features between them. In addition, existing graph-based DTI prediction methods often face the challenge of over-smoothing in graph neural networks (GNNs), which limits their prediction accuracy. To address these issues, we propose NASNet-DTI (Drug-target Interactions Based on Node Adaptation and Similarity Networks), a new framework designed to overcome these limitations. NASNet-DTI uses graph convolutional network to extract features from drug molecules and targets separately, and constructs heterogeneous networks to represent two types of nodes: drugs and targets. The edges in the network describe their multiple relationships: drug-drug, target-target, and drug-target. In the feature learning stage, NASNet-DTI adopts a node adaptive learning strategy to dynamically determine the optimal aggregation depth for each node. This ensures that each node can learn the most discriminative features, which effectively alleviates the over-smoothing problem and improves prediction accuracy. Experimental results show that NASNet-DTI significantly outperforms existing methods on multiple datasets, demonstrating its effectiveness and potential as a powerful tool to advance drug discovery and development.

Developing computational methods for single-cell drug response prediction deepens our understanding of tumor heterogeneity and uncovers resistance mechanisms critical to improving cancer therapy. However, current approaches struggle to fully capture intratumoral heterogeneity, as bulk RNA sequencing (bulk RNA-seq) obscures heterogeneity across individual cells, while single-cell RNA sequencing (scRNA-seq) remains constrained by limited throughput and high cost. Current approaches integrating bulk and scRNA-seq data frequently encounter batch effects, impairing robust knowledge transfer. Moreover, most existing methods overlook the role of intercellular interactions, treating cells as isolated entities. To overcome these limitations, we propose DAGFormer, a Graph-based Domain Adaptation framework that integrates bulk and scRNA-seq data for predicting single-cell drug responses. DAGFormer constructs cellular neighbor graphs using diverse topological strategies and employs Graph Domain Adaptation (GDA) to bridge graph-level distribution gaps between bulk and single-cell RNA-seq data. A dual-domain decoder further disentangles shared and modality-specific representations, preserving both general and unique biological signals. Benchmarking DAGFormer on ten independent scRNA-seq datasets demonstrated its superior performance compared to existing methods, underscoring its effectiveness and robustness in cancer drug response prediction.

Natural products, abundant in biological activities, are synthesized by bacteria during secondary metabolism, yielding a plethora of antibiotics, anticancer drugs, and other potential therapeutics. These compounds are directed by biosynthetic gene clusters (BGCs), which encode the necessary enzymes and regulatory elements for their production. With the rapid advancement of sequencing technology, the number of fully sequenced bacterial genomes has exponentially increased, providing researchers with an unprecedented opportunity to discover more BGCs. However, current identification methods are significantly constrained by their reliance on database quantity and predominantly utilize local features inherent in the BGC sequences. In this study, we introduce Biosynthetic Gene Clusters Catcher: GCN-BERT(BGCCGB), a deep learning model leveraging natural language processing for the identification of BGCs in bacterial genomes and classification of BGCs product types. Our approach represents BGCs as sequences represented by protein family (Pfam) identifiers and performs data augmentation on existing BGC data by introducing a synonym replacement method. Local and global features in BGCs sequences are then integrated by BERT and its self-attention mechanism, in conjunction with Graph Convolutional Networks (GCN). The experiment results indicate that BGCCGB effectively learns the features of BGCs, localizes BGCs in bacterial genomes, and predicts the product types of BGCs. Notably, BGCCGB demonstrates a significant performance improvement over previous methods, with an average precision 6.9% in BGC identification. These enhancements underscore the potential of BGCCGB in advancing the discovery and understanding of natural products synthesized by bacteria.

Protein methylation is one of the most important reversible post-translational modifications (PTMs), playing a vital role in the regulation of gene expression. Protein methylation sites serve as biomarkers in cardiovascular and pulmonary diseases, influencing various aspects of normal cell biology and pathogenesis. Nonetheless, the majority of existing computational methods for predictingprotein methylation sites(PMSP) have been constructed based on protein sequences, with few methods leveraging the topological information of proteins. To address this issue, we propose an innovative framework for predicting Methylation Sites using Graphs (GraphMethySite) that employs graph convolution network in conjunction with Bayesian Optimization (BO) to automatically discover the graphical structure surrounding a candidate site and improve the predictive accuracy. In order to extract the most optimal subgraphs associated with methylation sites, we extend GraphMethySite by coupling it with a hybrid Bayesian optimization (together named GraphMethySite$^+$) to determine and visualize the topological relevance among amino-acid residues. We evaluated our framework on two extended protein methylation datasets, and empirical results demonstrate that it outperforms existing state-of-the-art methylation prediction methods.

Transcription factors (TFs) are crucial proteins that regulate gene transcription by binding to specific sites on DNA, known as transcription factor binding sites (TFBSs). Identifying TFBSs enables the design of drugs to modulate gene expression, making it important for drug design and gene therapy. While deep learning-based methods have been proposed for predicting TFBSs, there is room for improvement. This study introduces TFBSnet, a novel deep learning-based technique that accurately predicts TFBSs by extracting diverse feature data from DNA sequences and utilizing a convolutional neural network (CNN) combined with SKNet. Experimental results show that TFBSnet outperforms existing methods. It also demonstrates accurate prediction of TF binding sites in human cells without label data and exceptional performance in predicting TFBSs in plant cells using 265 TFs in Arabidopsis. Ablation analysis highlights the integration of different features and advanced feature extraction by SKNet as contributors to TFBSnet's superior predictive capability.

Information freshness, characterized by Age of Information (AoI), is important for sensor applications involving timely status updates. In many cases, the wireless signals from one sensor can be received by multiple access points (APs). This article investigates the average AoI for cooperative APs (Co-APs), in which they can share information through a wired backbone network. We first study a basic backbone-assisted Co-AP system where APs share only decoded packets. Experimental results on software-defined radios (SDRs) indicate that Co-AP significantly improves the average AoI performance over a single-AP system. Next, we investigate an improved Co-AP system, called Soft-Co-AP. In addition to sharing decoded packets, Soft-Co-AP shares and collects soft information of packets that the APs fail to decode for further joint decoding. A critical issue in Soft-Co-AP is determining the number of quantization bits that represent the soft information (each soft bit) shared over the backbone. While more quantization bits per soft bit improves the joint decoding performance, it leads to higher backbone delay. We experimentally study the average AoI of Soft-Co-AP by evaluating the tradeoff between the backbone delay and the number of quantization bits. SDR experiments show that when the number of sensors is large, Soft-Co-AP further reduces the average AoI by 12% compared with Co-AP. Interestingly, good average AoI performance is usually achieved when the number of quantization bits per soft bit is neither too large nor too small.

Prokaryotic viruses, also known as bacteriophages, play crucial roles in regulating microbial communities and have the potential for phage therapy applications. Accurate prediction of phage-host interactions is essential for understanding the dynamics of these viruses and their impacts on bacterial populations. Numerous computational methods have been developed to tackle this challenging task. However, most existing prediction models can be constrained due to the substantial number of unknown interactions in comparison to the constrained diversity of available training data. To solve the problem, we introduce a model for prokaryotic virus host prediction with graph contrastive augmentation (PHPGCA). Specifically, we construct a comprehensive heterogeneous graph by integrating virus-virus protein similarity and virus-host DNA sequence similarity information. As the backbone encoder for learning node representations in the virus-prokaryote graph, we employ LGCN, a state-of-the-art graph embedding technique. Additionally, we apply graph contrastive learning to augment the node representations without the need for additional labels. We further conducted two case studies aimed at predicting the host range of multi-species phages, helping to understand the phage ecology and evolution.

Interactions between transcription factor and target gene form the main part of gene regulation network in human, which are still complicating factors in biological research. Specifically, for nearly half of those interactions recorded in established database, their interaction types are yet to be confirmed. Although several computational methods exist to predict gene interactions and their type, there is still no method available to predict them solely based on topology information. To this end, we proposed here a graph-based prediction model called KGE-TGI and trained in a multi-task learning manner on a knowledge graph that we specially constructed for this problem. The KGE-TGI model relies on topology information rather than being driven by gene expression data. In this paper, we formulate the task of predicting interaction types of transcript factor and target genes as a multi-label classification problem for link types on a heterogeneous graph, coupled with solving another link prediction problem that is inherently related. We constructed a ground truth dataset as benchmark and evaluated the proposed method on it. As a result of the 5-fold cross experiments, the proposed method achieved average AUC values of 0.9654 and 0.9339 in the tasks of link prediction and link type classification, respectively. In addition, the results of a series of comparison experiments also prove that the introduction of knowledge information significantly benefits to the prediction and that our methodology achieve state-of-the-art performance in this problem.

Transcription factors (TFs) are DNA binding proteins involved in the regulation of gene expression. They exist in all organisms and activate or repress transcription by binding to specific DNA sequences. Traditionally, TFs have been identified by experimental methods that are time-consuming and costly. In recent years, various computational methods have been developed to identify TF to overcome these limitations. However, there is a room for further improvement in the predictive performance of these tools in terms of accuracy. We report here a novel computational tool, TFnet, that provides accurate and comprehensive TF predictions from protein sequences. The accuracy of these predictions is substantially better than the results of the existing TF predictors and methods. Especially, it outperforms comparable methods significantly when sequence similarity to other known sequences in the database drops below 40%. Ablation tests reveal that the high predictive performance stems from innovative ways used in TFnet to derive sequence Position-Specific Scoring Matrix (PSSM) and encode inputs.

MOTIVATION: Interaction between transcription factor (TF) and its target genes establishes the knowledge foundation for biological researches in transcriptional regulation, the number of which is, however, still limited by biological techniques. Existing computational methods relevant to the prediction of TF-target interactions are mostly proposed for predicting binding sites, rather than directly predicting the interactions. To this end, we propose here a graph attention-based autoencoder model to predict TF-target gene interactions using the information of the known TF-target gene interaction network combined with two sequential and chemical gene characters, considering that the unobserved interactions between transcription factors and target genes can be predicted by learning the pattern of the known ones. To the best of our knowledge, the proposed model is the first attempt to solve this problem by learning patterns from the known TF-target gene interaction network. RESULTS: In this paper, we formulate the prediction task of TF-target gene interactions as a link prediction problem on a complex knowledge graph and propose a deep learning model called GraphTGI, which is composed of a graph attention-based encoder and a bilinear decoder. We evaluated the prediction performance of the proposed method on a real dataset, and the experimental results show that the proposed model yields outstanding performance with an average AUC value of 0.8864 +/- 0.0057 in the 5-fold cross-validation. It is anticipated that the GraphTGI model can effectively and efficiently predict TF-target gene interactions on a large scale. AVAILABILITY: Python code and the datasets used in our studies are made available at https://github.com/YanghanWu/GraphTGI.

Intrinsic disorder is common in proteins, plays important roles in protein functionality, and is commonly associated with various human diseases. To have an accurate tool for the annotation of intrinsic disorder in proteins, this paper proposes a novel algorithm, DeepCLD, for sequence-based prediction of intrinsically disordered proteins. This algorithm uses amino acid position specific scoring matrix (PSSM) to capture the intrinsic variability characteristic of sequence patterns, ResNet to preserve feature space structure, and bidirectional CudnnLSTM as recurrent layer to further improve the efficiency. Futhermore, DeepCLD also utilized the attention mechanism to solve the problem of gradient disappearing in deep network. Comparative analyses show that DeepCLD has faster training speed and higher prediction accuracy than comparable methods.

The broad implementation of cloud computing has brought a dramatic change to multiple industries, which derives from the development of the Internet-related technologies. This trend has enabled global enterprises to apply distributed computing techniques to reach many benefits. An effective risk management approach is required for service deliveries and a capacity planning is considered one of the convincing methods for financial industry. However, executing a capacity planning is still encountering a great challenge from bottlenecks of the Web server capacities. The unstable service demands often result in service delays, which embarrasses the competitivenesses of the enterprises. This paper addresses this issue and proposes an approach, named Efficiency-aware Cloud-based Workload Optimization (ECWO) Model, using greedy programming to predict server workloads of heterogeneous cloud computing in financial industry. The main algorithms used in the proposed model are Task Mapping Algorithm (TMA) and Efficiency-Aware Task Assignment (EATA) Algorithm. Our experimental evaluations have examined the performance of the proposed scheme.

Microarray technology enables the study of measuring gene expression levels for thousands of genes simultaneously. Cluster analysis of gene expression profiles has been applied for analyzing the function of gene because co-expressed genes are likely to share the same biological function. K-MEANS is one of well-known clustering methods. However, it requires a precise estimation of number of clusters and it has to assign all the genes into clusters. Other main problems are sensitive to the selection of an initial clustering and easily becoming trapped in a local minimum. We present a new clustering method for microarray gene data, called ppoCluster. It has two steps: 1) Estimate the number of clusters 2) Take sub-clusters resulting from the first step as input, and bridge a variation of traditional Particle Swarm Optimization (PSO) algorithm into K-MEANS for particles perform a parallel search for an optimal clustering. Our results indicate that ppoCluster is generally more accurate than K-MEANS and FKM. It also has better robustness for it is less sensitive to the initial randomly selected cluster centroids. And it outperforms comparable methods with fast convergence rate and low computation load.