Weizhong Lin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Weizhong Lin has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
Xuan Xiao
6 shared publicationsWangren Qiu
Affiliation School of Information Engineering, Jingdezhen Ceramic University, Jingdezhen, China
3 shared publicationsLiyi Yu
2 shared publicationsCunwen Liu
School of Information Engineering, Jingdezhen Ceramic University
1 shared publicationHanqing Zhang
Affiliation School of Information Engineering, Jingdezhen Ceramic University, Jingdezhen, China
1 shared publication- Hanqing Zhang
Jingdezhen ceramic Unicersity, Jingdezhen, Jiangxi, China
1 shared publication
Latest Publications
- Open Access
- 3 September 2026
- Published by Elsevier in Biochemistry and Biophysics Reports
- Vol. 47
KGTC-DDA:Drug-disease association prediction based on parallel graph isomorphism and transformer networks with cross-attention
- 3 September 2026
- Published by Elsevier in Biochemistry and Biophysics Reports
- Vol. 47
Open AccessKGTC-DDA:Drug-disease association prediction based on parallel graph isomorphism and transformer networks with cross-attention
- Open Access
- 11 August 2026
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 22 (8) , e1014552
Edge-aware GAT-based protein binding sites prediction
Accurate identification of protein binding sites is crucial for understanding biomolecular interaction mechanisms and for the rational design of drug targets. Traditional predictive methods often struggle to balance prediction accuracy with computational efficiency when capturing complex spatial conformations. To address this challenge, we propose an Edge-aware Graph Attention Network (Edge-aware GAT) model for the fine-grained prediction of binding sites across various biomolecules, including proteins, DNA/RNA, ions, ligands, and lipids. Our method constructs atom-level graphs and integrates multidimensional structural features, including geometric descriptors, DSSP-derived secondary structure, and relative solvent accessibility (RSA), to generate spatially aware embedding vectors. By incorporating interatomic distances and unit direction vectors as geometric edge features within the attention mechanism, the model enhances local structural representation without claiming strict SE(3)- or E(3)-equivariance. On benchmark datasets, our model achieves a ROC-AUC of 0.93 for protein-protein binding site prediction and shows competitive performance among the evaluated structure-based atomic-level models. The use of geometry-aware edge attention and residue-level attention pooling further improves both binding site localization and the capture of local structural details. Visualizations using PyMOL confirm the model’s practical utility and interpretability. To facilitate community access and application, we have deployed a publicly accessible web server at http://119.45.201.89:5000/ . In summary, our approach offers an efficient structure-based framework that balances prediction accuracy, generalization, and interpretability for identifying functional sites in proteins.
- 11 August 2026
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 22 (8) , e1014552
Open AccessEdge-aware GAT-based protein binding sites prediction
Accurate identification of protein binding sites is crucial for understanding biomolecular interaction mechanisms and for the rational design of drug targets. Traditional predictive methods often struggle to balance prediction accuracy with computational efficiency when capturing complex spatial conformations. To address this challenge, we propose an Edge-aware Graph Attention Network (Edge-aware GAT) model for the fine-grained prediction of binding sites across various biomolecules, including proteins, DNA/RNA, ions, ligands, and lipids. Our method constructs atom-level graphs and integrates multidimensional structural features, including geometric descriptors, DSSP-derived secondary structure, and relative solvent accessibility (RSA), to generate spatially aware embedding vectors. By incorporating interatomic distances and unit direction vectors as geometric edge features within the attention mechanism, the model enhances local structural representation without claiming strict SE(3)- or E(3)-equivariance. On benchmark datasets, our model achieves a ROC-AUC of 0.93 for protein-protein binding site prediction and shows competitive performance among the evaluated structure-based atomic-level models. The use of geometry-aware edge attention and residue-level attention pooling further improves both binding site localization and the capture of local structural details. Visualizations using PyMOL confirm the model’s practical utility and interpretability. To facilitate community access and application, we have deployed a publicly accessible web server at http://119.45.201.89:5000/ . In summary, our approach offers an efficient structure-based framework that balances prediction accuracy, generalization, and interpretability for identifying functional sites in proteins.
- Open Access
- 17 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 635-640
PPI-Guided Edge Predictor and Conditional Diffusion-Augmented GAT-GCN Framework for Accurate DNA-Binding Site Prediction
Accurate prediction of DNA binding sites is crucial for gene regulation and protein function analysis. However, the current task faces several challenges, namely severe class imbalance and insufficient integration of biological prior knowledge. We propose a PPI-based edge predictor and conditionally diffused enhanced GAT-GCN architecture to achieve high-precision DBS identification. Technically, we introduce a cross-scale transfer learning paradigm, where the edge predictor based on MLP transfers large-scale protein-protein interaction patterns to residue-residue contact prediction. To alleviate the extreme class imbalance problem, we design an enhanced conditional diffusion model, which has a specialized context encoder, synthesizing high-quality DBS samples through progressive denoising while retaining the global context of the protein. The hybrid GAT-GCN backbone network employs multi-head attention mechanisms and graph convolutional techniques to capture local residue interactions and global topological features in a hierarchical manner. A hybrid focal loss and class balance loss further optimizes learning on imbalanced distributions. Evaluation on benchmark datasets (DNA-573, DNA-646, and independent test set) shows that our model significantly outperforms state-of-the-art methods in F1, MCC, and AUC scores. This work advances computer technologies for bioinformatics: it improves graph construction through cross-scale knowledge transfer, upgrades data augmentation through the conditional diffusion model, and strengthens imbalanced learning through the design of mixed losses. The proposed framework provides a general GCN-diffusion paradigm for protein functional site prediction, contributing to the methodological progress of deep learning in computational biology.
- 17 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 635-640
Open AccessPPI-Guided Edge Predictor and Conditional Diffusion-Augmented GAT-GCN Framework for Accurate DNA-Binding Site Prediction
Accurate prediction of DNA binding sites is crucial for gene regulation and protein function analysis. However, the current task faces several challenges, namely severe class imbalance and insufficient integration of biological prior knowledge. We propose a PPI-based edge predictor and conditionally diffused enhanced GAT-GCN architecture to achieve high-precision DBS identification. Technically, we introduce a cross-scale transfer learning paradigm, where the edge predictor based on MLP transfers large-scale protein-protein interaction patterns to residue-residue contact prediction. To alleviate the extreme class imbalance problem, we design an enhanced conditional diffusion model, which has a specialized context encoder, synthesizing high-quality DBS samples through progressive denoising while retaining the global context of the protein. The hybrid GAT-GCN backbone network employs multi-head attention mechanisms and graph convolutional techniques to capture local residue interactions and global topological features in a hierarchical manner. A hybrid focal loss and class balance loss further optimizes learning on imbalanced distributions. Evaluation on benchmark datasets (DNA-573, DNA-646, and independent test set) shows that our model significantly outperforms state-of-the-art methods in F1, MCC, and AUC scores. This work advances computer technologies for bioinformatics: it improves graph construction through cross-scale knowledge transfer, upgrades data augmentation through the conditional diffusion model, and strengthens imbalanced learning through the design of mixed losses. The proposed framework provides a general GCN-diffusion paradigm for protein functional site prediction, contributing to the methodological progress of deep learning in computational biology.
- Open Access
- 11 April 2026
- Published by Elsevier in Computational Biology and Chemistry
- Vol. 124 (Pt 1) , 109053
A drug–target affinity prediction model integrating multimodal feature fusion and structural modeling
- 11 April 2026
- Published by Elsevier in Computational Biology and Chemistry
- Vol. 124 (Pt 1) , 109053
Open AccessA drug–target affinity prediction model integrating multimodal feature fusion and structural modeling
- Open Access
- 3 February 2026
- Published by Frontiers Media SA in Frontiers in Genetics
- Vol. 17, 1739720
TCN-5mC: a predictor of 5-methylcytosine sites based on multi-feature fusion and TCN-inspired block networks
Accurate identification of 5-methylcytosine (5 mC) sites in promoter regions is crucial for understanding epigenetic regulation, but experimental methods remain costly and time-consuming, highlighting the need for reliable computational prediction tools. While existing deep learning approaches, such as BiLSTM-based, Transformer-based, and pretrained language models, have advanced the field, opportunities remain for further improvements in capturing long-range dependencies and handling imbalanced genomic data. Here, we present TCN-5mC, a deep learning model that integrates Temporal Convolutional Networks (TCN) inspired block with Bidirectional Gated Recurrent Units (BiGRU) and employs hybrid One-Hot and Nucleotide Chemical Property feature encoding. This architecture is designed to more effectively model both extended sequence contexts and local patterns. The model achieves high predictive performance on imbalanced datasets from lung cancer cell lines, with AUC values of 0.967 and 0.989 on two independent test sets, outperforming existing methods in specificity, accuracy, MCC, and AUC. The model thus provides a robust, high-throughput computational tool for 5 mC site prediction, with promising potential for epigenetic research and biomarker discovery.
- 3 February 2026
- Published by Frontiers Media SA in Frontiers in Genetics
- Vol. 17, 1739720
Open AccessTCN-5mC: a predictor of 5-methylcytosine sites based on multi-feature fusion and TCN-inspired block networks
Accurate identification of 5-methylcytosine (5 mC) sites in promoter regions is crucial for understanding epigenetic regulation, but experimental methods remain costly and time-consuming, highlighting the need for reliable computational prediction tools. While existing deep learning approaches, such as BiLSTM-based, Transformer-based, and pretrained language models, have advanced the field, opportunities remain for further improvements in capturing long-range dependencies and handling imbalanced genomic data. Here, we present TCN-5mC, a deep learning model that integrates Temporal Convolutional Networks (TCN) inspired block with Bidirectional Gated Recurrent Units (BiGRU) and employs hybrid One-Hot and Nucleotide Chemical Property feature encoding. This architecture is designed to more effectively model both extended sequence contexts and local patterns. The model achieves high predictive performance on imbalanced datasets from lung cancer cell lines, with AUC values of 0.967 and 0.989 on two independent test sets, outperforming existing methods in specificity, accuracy, MCC, and AUC. The model thus provides a robust, high-throughput computational tool for 5 mC site prediction, with promising potential for epigenetic research and biomarker discovery.
- Open Access
- 13 May 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (5) , e0320817
iProtDNA-SMOTE: Enhancing protein-DNA binding sites prediction through imbalanced graph neural networks
Protein-DNA interactions play a crucial role in cellular biology, essential for maintaining life processes and regulating cellular functions. We propose a method called iProtDNA-SMOTE, which utilizes non-equilibrium graph neural networks along with pre-trained protein language models to predict DNA binding residues. This approach effectively addresses the class imbalance issue in predicting protein-DNA binding sites by leveraging unbalanced graph data, thus enhancing model’s generalization and specificity. We trained the model on two datasets, TR646 and TR573, and conducted a series of experiments to evaluate its performance. The model achieved AUC values of 0.850, 0.896, and 0.858 on the independent test datasets TE46, TE129, and TE181, respectively. These results indicate that iProtDNA-SMOTE outperforms existing methods in terms of accuracy and generalization for predicting DNA binding sites, offering reliable and effective predictions to minimize errors. The model has been thoroughly validated for its ability to predict protein-DNA binding sites with high reliability and precision. For the convenience of the scientific community, the benchmark datasets and codes are publicly available at https://github.com/primrosehry/iProtDNA-SMOTE.
- 13 May 2025
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (5) , e0320817
Open AccessiProtDNA-SMOTE: Enhancing protein-DNA binding sites prediction through imbalanced graph neural networks
Protein-DNA interactions play a crucial role in cellular biology, essential for maintaining life processes and regulating cellular functions. We propose a method called iProtDNA-SMOTE, which utilizes non-equilibrium graph neural networks along with pre-trained protein language models to predict DNA binding residues. This approach effectively addresses the class imbalance issue in predicting protein-DNA binding sites by leveraging unbalanced graph data, thus enhancing model’s generalization and specificity. We trained the model on two datasets, TR646 and TR573, and conducted a series of experiments to evaluate its performance. The model achieved AUC values of 0.850, 0.896, and 0.858 on the independent test datasets TE46, TE129, and TE181, respectively. These results indicate that iProtDNA-SMOTE outperforms existing methods in terms of accuracy and generalization for predicting DNA binding sites, offering reliable and effective predictions to minimize errors. The model has been thoroughly validated for its ability to predict protein-DNA binding sites with high reliability and precision. For the convenience of the scientific community, the benchmark datasets and codes are publicly available at https://github.com/primrosehry/iProtDNA-SMOTE.
- Open Access
- 1 February 2024
- Published by Bentham Science Publishers Ltd. in Current Pharmaceutical Design
- Vol. 30 (6) , 468-476
LSTM-SAGDTA: Predicting Drug-target Binding Affinity with an Attention Graph Neural Network and LSTM Approach
Introduction: Drug development is a challenging and costly process, yet it plays a crucial role in improving healthcare outcomes. Drug development requires extensive research and testing to meet the demands for economic efficiency, cures, and pain relief. Methods: Drug development is a vital research area that necessitates innovation and collaboration to achieve significant breakthroughs. Computer-aided drug design provides a promising avenue for drug discovery and development by reducing costs and improving the efficiency of drug design and testing. Results: In this study, a novel model, namely LSTM-SAGDTA, capable of accurately predicting drug-target binding affinity, was developed. We employed SeqVec for characterizing the protein and utilized the graph neural networks to capture information on drug molecules. By introducing self-attentive graph pooling, the model achieved greater accuracy and efficiency in predicting drug-target binding affinity. Conclusion: Moreover, LSTM-SAGDTA obtained superior accuracy over current state-of-the-art methods only by using less training time. The results of experiments suggest that this method represents a highprecision solution for the DTA predictor.
- 1 February 2024
- Published by Bentham Science Publishers Ltd. in Current Pharmaceutical Design
- Vol. 30 (6) , 468-476
Open AccessLSTM-SAGDTA: Predicting Drug-target Binding Affinity with an Attention Graph Neural Network and LSTM Approach
Introduction: Drug development is a challenging and costly process, yet it plays a crucial role in improving healthcare outcomes. Drug development requires extensive research and testing to meet the demands for economic efficiency, cures, and pain relief. Methods: Drug development is a vital research area that necessitates innovation and collaboration to achieve significant breakthroughs. Computer-aided drug design provides a promising avenue for drug discovery and development by reducing costs and improving the efficiency of drug design and testing. Results: In this study, a novel model, namely LSTM-SAGDTA, capable of accurately predicting drug-target binding affinity, was developed. We employed SeqVec for characterizing the protein and utilized the graph neural networks to capture information on drug molecules. By introducing self-attentive graph pooling, the model achieved greater accuracy and efficiency in predicting drug-target binding affinity. Conclusion: Moreover, LSTM-SAGDTA obtained superior accuracy over current state-of-the-art methods only by using less training time. The results of experiments suggest that this method represents a highprecision solution for the DTA predictor.
- Open Access
- 26 January 2024
- Published by Association for Computing Machinery (ACM)
- p. 414-419
Predicting Protein Subcellular Localization with Multi-label using GraphSAGE and Multi-head Attention Mechanism
Identifying where proteins are located within cells is crucial for understanding cellualr processes and functions, as well as for developing new therapeutic strategies. We have developed an innovative model that employs GraphSAGE and multi-head attention mechanism to predict the multi-label proteins localization. In our approach, protein sequences are represented as graphs and extracted features using GraphSAGE. Subsequently, the proteins features are fed to a multi-head attention model and a CNN block. Through rigorous evaluations, our model achieves a good performance in strict subset-accuracy with 0.47. Our approach has consistently outperformed existing state-of-the-art methods. All data and codes are available at https://github.com/sycamoretr/ProtLoc1.
- 26 January 2024
- Published by Association for Computing Machinery (ACM)
- p. 414-419
Open AccessPredicting Protein Subcellular Localization with Multi-label using GraphSAGE and Multi-head Attention Mechanism
Identifying where proteins are located within cells is crucial for understanding cellualr processes and functions, as well as for developing new therapeutic strategies. We have developed an innovative model that employs GraphSAGE and multi-head attention mechanism to predict the multi-label proteins localization. In our approach, protein sequences are represented as graphs and extracted features using GraphSAGE. Subsequently, the proteins features are fed to a multi-head attention model and a CNN block. Through rigorous evaluations, our model achieves a good performance in strict subset-accuracy with 0.47. Our approach has consistently outperformed existing state-of-the-art methods. All data and codes are available at https://github.com/sycamoretr/ProtLoc1.
- Open Access
- 24 February 2023
- Published by MDPI AG in International Journal of Molecular Sciences
- Vol. 24 (5) , 4500
MSEDDI: Multi-Scale Embedding for Predicting Drug—Drug Interaction Events
A norm in modern medicine is to prescribe polypharmacy to treat disease. The core concern with the co-administration of drugs is that it may produce adverse drug—drug interaction (DDI), which can cause unexpected bodily injury. Therefore, it is essential to identify potential DDI. Most existing methods in silico only judge whether two drugs interact, ignoring the importance of interaction events to study the mechanism implied in combination drugs. In this work, we propose a deep learning framework named MSEDDI that comprehensively considers multi-scale embedding representations of the drug for predicting drug—drug interaction events. In MSEDDI, we design three-channel networks to process biomedical network-based knowledge graph embedding, SMILES sequence-based notation embedding, and molecular graph-based chemical structure embedding, respectively. Finally, we fuse three heterogeneous features from channel outputs through a self-attention mechanism and feed them to the linear layer predictor. In the experimental section, we evaluate the performance of all methods on two different prediction tasks on two datasets. The results show that MSEDDI outperforms other state-of-the-art baselines. Moreover, we also reveal the stable performance of our model in a broader sample set via case studies.
- 24 February 2023
- Published by MDPI AG in International Journal of Molecular Sciences
- Vol. 24 (5) , 4500
Open AccessMSEDDI: Multi-Scale Embedding for Predicting Drug—Drug Interaction Events
A norm in modern medicine is to prescribe polypharmacy to treat disease. The core concern with the co-administration of drugs is that it may produce adverse drug—drug interaction (DDI), which can cause unexpected bodily injury. Therefore, it is essential to identify potential DDI. Most existing methods in silico only judge whether two drugs interact, ignoring the importance of interaction events to study the mechanism implied in combination drugs. In this work, we propose a deep learning framework named MSEDDI that comprehensively considers multi-scale embedding representations of the drug for predicting drug—drug interaction events. In MSEDDI, we design three-channel networks to process biomedical network-based knowledge graph embedding, SMILES sequence-based notation embedding, and molecular graph-based chemical structure embedding, respectively. Finally, we fuse three heterogeneous features from channel outputs through a self-attention mechanism and feed them to the linear layer predictor. In the experimental section, we evaluate the performance of all methods on two different prediction tasks on two datasets. The results show that MSEDDI outperforms other state-of-the-art baselines. Moreover, we also reveal the stable performance of our model in a broader sample set via case studies.