AttentionMGT-DTA: A multi-modal drug-target affinity prediction using graph transformer and attention mechanism

Neural Networks, Journal Year: 2023, Volume and Issue: 169, P. 623 - 636

Published: Nov. 11, 2023

The accurate prediction of drug-target affinity (DTA) is a crucial step in drug discovery and design. Traditional experiments are very expensive time-consuming. Recently, deep learning methods have achieved notable performance improvements DTA prediction. However, one challenge for learning-based models appropriate representations drugs targets, especially the lack effective exploration target representations. Another how to comprehensively capture interaction information between different instances, which also important predicting DTA. In this study, we propose AttentionMGT-DTA, multi-modal attention-based model AttentionMGT-DTA represents targets by molecular graph binding pocket graph, respectively. Two attention mechanisms adopted integrate interact protein modalities pairs. experimental results showed that our proposed outperformed state-of-the-art baselines on two benchmark datasets. addition, had high interpretability modeling strength atoms residues. Our code available at https://github.com/JK-Liu7/AttentionMGT-DTA.

Language: Английский

---

Regulation-aware graph learning for drug repositioning over heterogeneous biological network

Information Sciences, Journal Year: 2024, Volume and Issue: 686, P. 121360 - 121360

Published: Aug. 22, 2024

Language: Английский

---

AMDGT: Attention aware multi-modal fusion using a dual graph transformer for drug–disease associations prediction

Knowledge-Based Systems, Journal Year: 2023, Volume and Issue: 284, P. 111329 - 111329

Published: Dec. 28, 2023

Identification of new indications for existing drugs is crucial through the various stages drug discovery. Computational methods are valuable in establishing meaningful associations between and diseases. However, most predict drug-disease based solely on similarity data, neglecting biological chemical information. These often use basic concatenation to integrate information from different modalities, limiting their ability capture features a comprehensive in-depth perspective. Therefore, novel multimodal framework called AMDGT was proposed dual-graph transformer modules. By combining data complex biochemical information, understands feature fusion diseases effectively comprehensively with an attention-aware modality interaction architecture. Extensive experimental results indicate that surpasses state-of-the-art real-world datasets. Moreover, case molecular docking studies demonstrated effective tool repositioning. Our code available at GitHub: https://github.com/JK-Liu7/AMDGT.

Language: Английский

---

Identifying virulence factors using graph transformer autoencoder with ESMFold-predicted structures

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 170, P. 108062 - 108062

Published: Jan. 30, 2024

Language: Английский

---

Drug repositioning by multi-aspect heterogeneous graph contrastive learning and positive-fusion negative sampling strategy

Information Fusion, Journal Year: 2024, Volume and Issue: 112, P. 102563 - 102563

Published: July 6, 2024

Drug repositioning (DR) is a promising approach for identifying novel indications of existing drugs. Computational methods drug have been recognised as effective ways to discover the associations between drugs and diseases. However, most computational DR ignore significance heterogeneous graph augmentation when conducting contrastive learning, which plays critical role in improving generalisation robustness. The high-order similarity information from multiple data sources still under-explored. Furthermore, only limited number can effectively screen informative negative samples model training. To address these limitations, we propose method called DRMAHGC that employs multi-aspect learning predict drug-disease (DDAs). First, features were generated network using graph-masked autoencoder. Then, with structure- metapath-level was employed enhance semantic comprehension learn expressive representations. Subsequently, positive-fusion sampling strategy exploited synthesise sample embeddings train classifier predicting DDAs. Extensive results on three benchmark datasets indicate significantly consistently outperformed state-of-the-art task. Moreover, case study two common diseases further demonstrates its effectiveness provides insights into

Language: Английский

---

BiRNN-DDI: A Drug-Drug Interaction Event Type Prediction Model Based on Bidirectional Recurrent Neural Network and Graph2Seq Representation

Journal of Computational Biology, Journal Year: 2024, Volume and Issue: unknown

Published: July 25, 2024

Research on drug-drug interaction (DDI) prediction, particularly in identifying DDI event types, is crucial for understanding adverse drug reactions and combinations. This work introduces a Bidirectional Recurrent Neural Network model type prediction (BiRNN-DDI), which simultaneously considers structural relationships contextual information. Our BiRNN-DDI constructs feature graphs to mine relationships. For information, it transforms into sequences employs two-channel structure, integrating BiRNN, obtain representations of pairs. The model's effectiveness demonstrated through comparisons with state-of-the-art models two event-type benchmarks. Extensive experimental results reveal that surpasses other accuracy, AUPR, AUC, F1 score, Precision, Recall metrics both small large datasets. Additionally, our exhibits lower parameter space, indicating more efficient learning potential types.

Language: Английский

---

MFD–GDrug: multimodal feature fusion-based deep learning for GPCR–drug interaction prediction

Methods, Journal Year: 2024, Volume and Issue: 223, P. 75 - 82

Published: Jan. 28, 2024

Language: Английский

---

Drug side effects prediction via cross attention learning and feature aggregation

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 248, P. 123346 - 123346

Published: Jan. 30, 2024

Language: Английский

---

Identification of drug-side effect association via correntropy-loss based matrix factorization with neural tangent kernel

Methods, Journal Year: 2023, Volume and Issue: 219, P. 73 - 81

Published: Sept. 30, 2023

Language: Английский

---

Role of PLA2R domain antibodies and epitope spreading in risk stratification and prediction of proteinuria remission in primary membranous nephropathy

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Feb. 26, 2025

Language: Английский

---