Protein phosphorylation governs cellular signaling, making accurate identification of kinase-specific sites essential for understanding regulatory and disease mechanisms. Although computational approaches have shown promise in inferring kinase specificity, most existing methods primarily rely on local sequence patterns and remain limited in their ability to capture broader contextual information. More critically, experimentally validated kinase-substrate data are inherently scarce and highly imbalanced across kinase groups, substantially restricting the generalization performance of purely discriminative models, especially for underrepresented kinases. To address these challenges, we propose KSDiffusion, a unified framework for kinase-specific phosphorylation site prediction explicitly designed for data-limited and imbalanced regimes. KSDiffusion integrates a protein language model with task-aware conditional diffusion-based generative modeling. Specifically, an ESM-2-based encoder is employed to extract context-aware peptide representations enriched with evolutionary and structural information, while supervised contrastive learning further enhances kinase-specific discriminability in the embedding space. To alleviate data scarcity for rare kinase groups, we introduce a conditional diffusion model, termed KS-DiT, which generates biologically plausible and kinase-consistent synthetic representations that directly support downstream prediction. Comprehensive experiments across kinase groups spanning low-, medium-, and large-data regimes demonstrate that KSDiffusion consistently outperforms representative baseline methods. In particular, substantial improvements are achieved for data-scarce kinase groups, with AUC gains of up to $\sim $15%, while maintaining competitive performance when sufficient training data are available. These results underscore the regime-dependent effectiveness of conditional diffusion-based augmentation and highlight the value of integrating protein language models with task-aware generative modeling for robust kinase-specific phosphorylation site prediction under realistic data constraints.

Protein phosphorylation is a critical posttranslational modification involved in cell signaling and metabolic regulation, and phosphorylation dysregulation is closely associated with various diseases. However, studies on phosphorylation site-disease associations remain scarce, and existing methods often fail to fully leverage complementary information across heterogeneous data sources, resulting in suboptimal predictive performance. We present CMV-GLA (Contrastive Multi-View Graph Layer Attention Network), a novel framework integrating protein sequences, disease semantics, and known associations into three complementary graph views (site similarity, disease similarity, association network). Built on a GAT backbone, CMV-GLA employs layer attention to mitigate over-smoothing and contrastive learning to maximize cross-view node embedding consistency while minimizing inter-node confusion, enhancing discriminative power. Specifically, the contrastive learning module encourages consistent representations of the same node across different graph views while explicitly separating embeddings of unrelated nodes, leading to more distinguishable and robust feature representations. Evaluated on benchmark data, CMV-GLA significantly outperforms state-of-the-art methods in AUC and AUPRC. Ablation studies confirm the critical roles of multi-view fusion and the contrastive module. Case studies demonstrate high-confidence, literature-supported predictions, highlighting CMV-GLA's utility for elucidating phosphorylation mediated mechanisms and guiding therapeutic discovery. Code and dataset available at https://github.com/ljr078/CMV-GLA.

Abstract Blockchain technology has rapidly evolved due to its decentralization and traceability; however, its inherent anonymity also facilitates phishing fraud, resulting in substantial financial losses. Existing graph-based phishing detection methods often underexploit temporal transaction dynamics and rely on relatively shallow spatio-temporal feature fusion. To address these limitations, we propose a Temporal transaction graph attention network (TTGAN) for phishing account identification. First, an attributed transaction multigraph is constructed to model transactional interactions, explicitly preserving duplicate edges with associated timestamps and transaction values. Second, a temporal random walk module is designed to capture temporal dependencies, where the maximum walk length is adaptively determined by the transaction graph scale, and node sampling follows a time-biased exponential probability scheme; the temporal decay effect is implicitly incorporated through this design. In parallel, a graph attention mechanism module learns spatial representations by jointly modeling node and edge attributes. Finally, temporal and spatial features are fused via concatenation, followed by average pooling and a fully connected layer for account classification. Experiments conducted on four real-world Ethereum datasets demonstrate that TTGAN achieves a precision of 91.23%–96.39%, a recall of 94.41%–95.99%, and an F1-score of 92.20%–96.14%, significantly outperforming state-of-the-art methods, including Graph Attention Networks (GAT), temporal transaction subgraph network, and Graphormer.

Few-shot action recognition seeks to recognize novel actions with limited labeled examples. While dual-modal approaches incorporating video and textual modalities offer enhanced semantic context, existing methods often rely on naive feature fusion strategies, failing to capture deep semantic correlations across modalities and limiting generalization. We propose DPCA-Net, a dual-modal metric learning framework that constructs a unified dual-prototype consistency alignment space. DPCA-Net explicitly models distributional, structural, and metric consistency across modalities to enhance prototype quality and similarity estimation. It integrates three core components: (1) Frame-wise Text-guided Modeling (FTM), which uses conditional prompt learning to embed video frame-level visual features into the textual space, achieving structural consistency; (2) Dual-Modal Metric Learning via dual-path Dynamic Time Warping (Dual-DTW), jointly aligning visual and cross-modal prototypes to ensure metric consistency; and (3) Distribution Consistency Mapping (DCM), which leverages Maximum Mean Discrepancy and cosine similarity to align support-query distributions and reinforce representation robustness. Extensive experiments on three benchmark datasets show that DPCA-Net consistently outperforms prior methods. It surpasses CLIP-FSAR by 1.3%–2.7%, achieving 89.7% (1-shot) on Kinetics and 99.12% (5-shot) on UCF-101. These results highlight the effectiveness of consistency-driven prototype alignment for robust and generalizable cross-modal few-shot action recognition.

Inter-protein post-translational modification (PTM) crosstalk critically regulates protein function and cellular signaling, yet its computational prediction remains challenging. Existing methods largely rely on hand-crafted features and struggle to integrate sequence and structural information from interacting proteins. To address this, we propose ProXTalk, a novel dual-stream framework that leverages protein large language models (pLLMs) to model bidirectional interactions between PTM sites across different proteins. In the Inter _3549 benchmark, ProXTalk achieves a state-of-the-art AUC of 0.906, surpassing the previous best method by 3.4 %. Comprehensive ablation studies verify the critical contributions of the pLLM -derived representations and the symmetric cross-attention mechanism. Code and data are available at: https://github.com/sisiou/ProXTalk.

Heart sounds reflect the function of heart valves and are important for the diagnosis of heart-related diseases. Automated heart sound diagnosis plays an key role in the early detection of cardiovascular diseases. In this paper, a new deep learning-based detection model (2D ViT-1D CRNN) for heart sound signals is designed, which combines one-dimensional time-domain and two-dimensional time-frequency domain features. In this model, a 1D CNN and a BiLSTM are combined into a 1D CRNN module to extract 1D temporal features from the original PCG signals, while 2D time-frequency features are extracted using a 2D ViT module. The classification results are calculated by softmax function after connecting the outputs of the two modules. To validate the effectiveness of the proposed 2D ViT-1D CRNN model, it is applied to two public datasets with different classification tasks. Compared with existing SOTA methods, our proposed method performs best on both datasets, which suggests that the method in this paper can be applied to assist diagnosis of cardiovascular diseases.

Automatic heart sound diagnosis plays an important role in the early detection of cardiovascular diseases. Phonocardiogram (PCG) signals are often used in this field f or its low cost and non-invasive advantages. In this paper, we design a new time and time-frequency features integrated CNN (TTFI-CNN) model for heart sound signals detection. In the model, a 1D CNN and a BiLSTM are combined into 1D CRNN module to extract temporal features from the original PCG signal, and a 2D CNN module is applied to capture high-level features from time-frequency domain MFCCs inputs. The outputs of the two modules are recalibrated by attention mechanism to selectively emphasize informative features and suppress less useful ones. To verify the proposed TTFI-CNN model, it is applied to two public datasets with different classification t asks (binary and multiclassification). The TTFI-CNN model has achieved 97.15% accuracy, 97.13% sensitivity, and 97.17% specificity on physionet/cinc database, and obtained 3.31 and 2.61 precision scores on the PASCAL database A and B, respectively. Compared with the previous state-of-the-art methods, the TTFI-CNN performs best on all the above metrics. https://github.com/XxxNnnSssWww/TTFI-CNN.

The subcellular localization of proteins is very important for further understanding their functions. This paper proposes a deep learning method called PBLoc to predict the protein subcellular localization from sequence information only. In the PBLoc method, the pre-trained protein model–ProtBert, is utilized to extract the features of the protein sequence, followed by a bidirectional GRU model to predict subcellular localization. In order to capture sequence features more accurately, an attention mechanism is also used in the model. To verify the performance of our proposed PBLoc method, it is applied to DeepLoc benchmark dataset. For comparison, other 8 SOTA algorithms are also executed. The results show that PBLoc without using any evolution feature outperforms the other compared algorithms.

Protein secondary structure prediction is still a challenging task in bioinformatics, especially for 8-state (Q8) classification. To address this problem, we have proposed a deep learning based model by integrating graph convolutional network(GCN) and bidirectional long short-term memory (Bi-LSTM) network in this paper. In the model, GCN is utilized to synthesize the information of amino acids and their interactions, while Bi-LSTM has strong ability to capture the long-range dependencies of amino acids. For sequence representation, a new protein embedding derived by ProtTrans is used instead of the traditional amino acid one-hot encoding, together with evolutionary features of PSSM and HHM profiles. Amino acid contact potential derived from SPOTContact-Helical is used to construct amino acid graph. To verify the effectiveness of our proposed model, it is applied to several benchmark datasets, and obtained 78.05%, 76.81% 72.84%, 74.46% and 76.04% Q8 accuracy on CASP10, CASP11, CASP12, CB513 and TS115 datasets, respectively. Compared with 8 state-of-the-art competitions, our model obtained the best performance in most of datasets.

Abstract This paper studies the couplet generation model which automatically generates the second line of a couplet by giving the first line. Unlike other sequence generation problems, couplet generation not only considers the sequential context within a sentence line but also emphasizes the relationships between the corresponding words of first and second lines. Therefore, a trapezoidal context character embedding the vector model has been developed firstly, which considers the ‘sequence context’ and the ‘corresponding word context’ simultaneously. Afterwards, we chose the typical encoder–decoder framework to solve the sequence–sequence problems, of which the encoder and decoder are used by bi-directional GRU and GRU, respectively. In order to further increase the semantic consistency of the first and second lines of couplets, the pre-trained sentence vector of the first line is added to the attention mechanism in the model. To verify the effectiveness of the method, it is applied to the real data set. Experimental results show that our proposed model can compete with the up-to-date methods, and both adding sentence vectors to attention and using trapezoidal context character vectors can improve the effectiveness of the algorithm.

The study on the prediction of mine pressure, while exploiting in coal mine, is a critical and technical guarantee for coal mine safety and production. In this paper, primarily due to the actual demand for the prediction of mine pressure, a practical prediction model Mine Pressure Prediction (MPP) was proposed based on fuzzy cognitive maps (FCMs). The Real Coded Genetic Algorithm (RCGA) was proposed to solve the problem by introducing the weight regularization and dropout regularization. A numerical example involving in-situ monitoring data is studied. Mean Square Error (MSE) and fitness function were used to evaluate the applicability of MPP model which is trained by RCGA, Regularization Genetic Algorithm (RGA) and Weight and Dropout RGA optimization algorithms. The numerical results demonstrate that the proposed Weight and Dropout RGA is better than the other two algorithms, and realizing the requirement for prediction of mine pressure in the coal mine production.

The surprisingly popular decision in social science fields is a wisdom of the crowd technique that taps into the expert minority opinion within a crowd, which has been demonstrated to be remarkably effective for multiple questions. Most of the existing PSO variants construct the exemplars by solely using fitness, which could be viewed as the democratic approaches or methods. However, the democratic methods tend to highlight the most popular opinion, not necessarily the most correct, which might lead the population into a local trapping region in the scenarios of swarm intelligent computing and evolutionary computation. This paper proposes a method to implement the surprisingly popular decision in PSO to facilitate the exemplar construction, cooperating with the dynamic topology maintenance. The proposed PSO variant is called the Surprisingly Popular Algorithm-based Comprehensive Adaptive Topology Learning Particle Swarm Optimization (SPA-CatlePSO). By using the dynamic topological connection and surprisingly popular decision strategy, the proposed SPA-CatlePSO could adjust the degree of small world topology, mimicking the mechanism of knowledge conversion in the crowd, and guide the direction of the exploitation by constructing exemplars with the largest surprisingly popular degree. We evaluate the proposed SPA-CatlePSO on the full CEC2014 benchmark suite and compare its validity with OLPSO, TSLPSO, ASDPSO, HCLPSO, OptBees and L-shade. The experimental results show that the SPA-CatlePSO algorithm is competitive with the most advanced swarm-based intelligent algorithms.

BACKGROUND: RNAs have been found to carry diverse functionalities in nature. Inferring the similarity between two given RNAs is a fundamental step to understand and interpret their functional relationship. The majority of functional RNAs show conserved secondary structures, rather than sequence conservation. Those algorithms relying on sequence-based features usually have limitations in their prediction performance. Hence, integrating RNA structure features is very critical for RNA analysis. Existing algorithms mainly fall into two categories: alignment-based and alignment-free. The alignment-free algorithms of RNA comparison usually have lower time complexity than alignment-based algorithms. RESULTS: An alignment-free RNA comparison algorithm was proposed, in which novel numerical representations RNA-TVcurve (triple vector curve representation) of RNA sequence and corresponding secondary structure features are provided. Then a multi-scale similarity score of two given RNAs was designed based on wavelet decomposition of their numerical representation. In support of RNA mutation and phylogenetic analysis, a web server (RNA-TVcurve) was designed based on this alignment-free RNA comparison algorithm. It provides three functional modules: 1) visualization of numerical representation of RNA secondary structure; 2) detection of single-point mutation based on secondary structure; and 3) comparison of pairwise and multiple RNA secondary structures. The inputs of the web server require RNA primary sequences, while corresponding secondary structures are optional. For the primary sequences alone, the web server can compute the secondary structures using free energy minimization algorithm in terms of RNAfold tool from Vienna RNA package. CONCLUSION: RNA-TVcurve is the first integrated web server, based on an alignment-free method, to deliver a suite of RNA analysis functions, including visualization, mutation analysis and multiple RNAs structure comparison. The comparison results with two popular RNA comparison tools, RNApdist and RNAdistance, showcased that RNA-TVcurve can efficiently capture subtle relationships among RNAs for mutation detection and non-coding RNA classification. All the relevant results were shown in an intuitive graphical manner, and can be freely downloaded from this server. RNA-TVcurve, along with test examples and detailed documents, are available at: http://ml.jlu.edu.cn/tvcurve/ .

Based on an effective clustering algorithm-Affinity Propagation (AP)-we present in this paper a novel semisupervised text clustering algorithm, called Seeds Affinity Propagation (SAP). There are two main contributions in our approach: 1) a new similarity metric that captures the structural information of texts, and 2) a novel seed construction method to improve the semisupervised clustering process. To study the performance of the new algorithm, we applied it to the benchmark data set Reuters-21578 and compared it to two state-of-the-art clustering algorithms, namely, k-means algorithm and the original AP algorithm. Furthermore, we have analyzed the individual impact of the two proposed contributions. Results show that the proposed similarity metric is more effective in text clustering (F-measures ca. 21 percent higher than in the AP algorithm) and the proposed semisupervised strategy achieves both better clustering results and faster convergence (using only 76 percent iterations of the original AP). The complete SAP algorithm obtains higher F-measure (ca. 40 percent improvement over k-means and AP) and lower entropy (ca. 28 percent decrease over k-means and AP), improves significantly clustering execution time (20 times faster) in respect that k-means, and provides enhanced robustness compared with all other methods.

Affinity propagation is an impressive clustering algorithm which was published in Science, 2007. However, the original algorithm couldn't cope with part known data directly. Focusing on this issue, a semi-supervised scheme called incremental affinity propagation clustering is proposed in the paper. In the scheme, the pre-known information is represented by adjusting similarity matrix. Moreover, an incremental study is applied to amplify the prior knowledge. To examine the effectiveness of the method, we concentrate it to text clustering problem and describe the specific method accordingly. The method is applied to the benchmark data set Reuters-21578. Numerical results show that the proposed method performs very well on the data set and has most advantages over two other commonly used clustering methods.

Principal component analysis (PCA) is a favorite multivariate statistical method for image enhancement and compression. However, it is well known that the classical PCA is sensitive to outliers and missing data. Fortunately, fuzzy statistics is an effective theory for processing these kinds of data. Fuzziness and randomicity are just the important characteristics of the data of remote-sensing images. Therefore, by introducing fuzzy statistics variables into classical PCA methods, a novel method for multispectral image processing called fuzzy-statistics-based PCA (FS-PCA) is proposed in this paper. To verify our proposed method, both the classical PCA and the FS-PCA are applied to the multispectral Landsat ETM+ data for image enhancement. The experimental results show that the differences among surface characteristics are expanded sufficiently and that the accuracy of surface feature recognition is improved greatly.

Inspired by the idea of genetic algorithm, we propose two hybrid evolutionary algorithms based on PSO and GA methods through crossing over the PSO and GA algorithms. The main ideas of the two proposed methods are to integrate PSO and GA methods in parallel and series forms respectively. Simulations for a series of benchmark test functions show that both of the two proposed methods possess better ability to find the global optimum than that of the standard PSO algorithm.