Relation-aware Heterogeneous Graph Transformer based drug repurposing

Mei, Xin; Cai, Xiaoyan; Yang, Linlin; Wang, Nanxin

doi:10.1016/j.eswa.2021.116165

Cited by 26 publications

(11 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The material basis and mechanism of action of the formulation were analyzed and predicted by constructing the model. Such a graph neural network model is essentially a deep learning approach that makes full use of the interaction relationships and structural information in the CTP heterogeneous network [12], but also draws on the networked, systematic and holistic ideas of network pharmacology [51], and allows the prediction results to be validated to demonstrate reliability.…”

Section: Discussionmentioning

confidence: 99%

“…Both the GraphSAGE model and the GAT model used in this study follow such a process of node updating. On the other hand, our multi-label classification model and link prediction model have been designed to be endto-end models, so the features of the nodes are continuously updated during the iterative process, which is one of the characteristics that distinguishes deep learning from traditional machine learning [12]. In addition, forward propagation yields the output, calculates the loss using the loss function and back propagates to optimize the parameters, thus ensuring that the performance of the model is as optimal as possible [57], which is also a powerful aspect of deep learning.…”

Section: Discussionmentioning

confidence: 99%

“…Graph neural network (GNN) is a deep learning model that applies deep learning to graph structures and can model graph network data consisting of nodes and edges [11]. Specifically, node-level, edge-level and graph-level modelling can be accomplished [12]. Drawing on the idea of GNN for modelling graph data, we apply GNN to the CTP complex heterogeneous network of Chinese medicine.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Application of a Complex Network Modeling Approach to Explore the Material Basis and Mechanisms of Traditional Chinese Medicine: A Case Study of Xuefu Zhuyu Decoction for the Treatment of Two Types of Angina Pectoris

Zhang

et al. 2022

IEEE Access

View full text Add to dashboard Cite

The complex processes of action of Chinese medicines and their formulations can be studied in terms of complex heterogeneous networks, however multiple diseases are different from one disease. In this study, we proposed a graph neural network-based modelling strategy to explore the material basis and mechanisms of the formulae under multiple disease conditions, exemplified by Xuefu Zhuyu Decoction for stable angina pectoris and unstable angina pectoris. Our modelling approach consists of two parts, a feature part and a model part. In the feature part, we calculated four similarities and their combinations in the similarity network as features of nodes of the heterogeneous network, using a variety of algorithms. In the model part, both the multi-label classification model for compounds and the link prediction model for compounds-targets were performed in six model comparison experiments and four partial ablation experiments. We confirmed that the GraphSAGEbased model was the best performing multi-label classification model while the GAT-based model was the best performing link prediction model, and that the full model with various similarity contents was optimal. Good generalization ability of the model was demonstrated through generalization ability experiments. In the case study, the model was used to predict potential compound-target interactions for marker targets and the reliability of the predictions was demonstrated by molecular docking and literature mining. In conclusion, the model successfully predicted the known material basis and mechanisms and also predicted unknown compound-marker target interactions. The proposed modelling approach can be extended to the study of other formulations. INDEX TERMSHeterogeneous network; material basis; mechanism; multi-label classification; link prediction I. INTRODUCTION Chinese herbs and their constituent formulas are widely used in clinical practice to treat various types of diseases[1, 2], yet the scientific and systematic elucidation of the material basis and mechanism of action of Chinese herbs and formulas has been a challenge that This article has been accepted for publication in IEEE Access.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Application of a Complex Network Modeling Approach to Explore the Material Basis and Mechanisms of Traditional Chinese Medicine: A Case Study of Xuefu Zhuyu Decoction for the Treatment of Two Types of Angina Pectoris

Zhang

et al. 2022

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Another popular paradigm of KG embedding is via graph neural networks, such as Graph Convolutional Networks (GCN) [32] and Graph Attention Networks (GAT) [33]. Recently, graph embedding based on Transfomers [34] have been proposed such as HGT [35] and RHGT [36]. These methods learn entity embeddings via information propagation between nodes in a graph.…”

Section: Knowledge Base Embedding Methodsmentioning

confidence: 99%

TransAlign: Fully Automatic and Effective Entity Alignment for Knowledge Graphs

Zhang¹,

Zhao²,

Trisedya³

et al. 2022

Preprint

View full text Add to dashboard Cite

The task of entity alignment between knowledge graphs (KGs) aims to identify every pair of entities from two different KGs that represent the same entity. Many machine learning-based methods have been proposed for this task. However, to our best knowledge, existing methods all require manually crafted seed alignments, which are expensive to obtain. In this paper, we propose the first fully automatic alignment method named TransAlign, which does not require any manually crafted seed alignments. Specifically, for predicate embeddings, TransAlign constructs a predicate-proximity-graph to automatically capture the similarity between predicates across two KGs by learning the attention of entity types. For entity embeddings, TransAlign first computes the entity embeddings of each KG independently using TransE, and then shifts the two KGs' entity embeddings into the same vector space by computing the similarity between entities based on their attributes. Thus, both predicate alignment and entity alignment can be done without manually crafted seed alignments. TransAlign is not only fully automatic, but also highly effective. Experiments using real-world KGs show that TransAlign improves the accuracy of entity alignment significantly compared to state-of-the-art methods.

show abstract

“…Many miRNAs regulate many genes and in turn afect many pathways. Tis complex process can be graphically represented by the miRNA-target gene-pathway heterogeneous network (MTP), which we refer to as the miRNA regulatory network [21]. On the one hand, the constructed MTP networks can be used to construct miRNA similarity networks and fnd hub miRNAs by calculating the similarity of miRNA actions, i.e., network analysis.…”

Section: Introductionmentioning

confidence: 99%

A Complex Heterogeneous Network Model of Disease Regulated by Noncoding RNAs: A Case Study of Unstable Angina Pectoris

Dan

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

MicroRNAs (miRNAs) are important types of noncoding RNAs, and there is a lack of holistic and systematic understanding of the functions they play in disease. We proposed a research strategy, including two parts network analysis and network modelling, to analyze, model, and predict the regulatory network of miRNAs from a network perspective, using unstable angina pectoris as an example. In the network analysis section, we proposed the WGCNA & SimCluster method using both correlation and similarity to find hub miRNAs, and validation on two datasets showed better results than the methods using correlation or similarity alone. In the network modelling section, we used six knowledge graph or graph neural network models for link prediction of three types of edges and multilabel classification of two types of nodes. Comparative experiments showed that the RotatE model was a good model for link prediction, while the RGCN model was the best model for multilabel classification. Potential target genes were predicted for hub miRNAs and validation of hub miRNA-target gene interactions, target genes as biomarkers and target gene functions were performed using a three-step validation approach. In conclusion, our study provides a new strategy to analyze and model miRNA regulatory networks.

show abstract

Relation-aware Heterogeneous Graph Transformer based drug repurposing

Cited by 26 publications

References 22 publications

Application of a Complex Network Modeling Approach to Explore the Material Basis and Mechanisms of Traditional Chinese Medicine: A Case Study of Xuefu Zhuyu Decoction for the Treatment of Two Types of Angina Pectoris

Application of a Complex Network Modeling Approach to Explore the Material Basis and Mechanisms of Traditional Chinese Medicine: A Case Study of Xuefu Zhuyu Decoction for the Treatment of Two Types of Angina Pectoris

TransAlign: Fully Automatic and Effective Entity Alignment for Knowledge Graphs

A Complex Heterogeneous Network Model of Disease Regulated by Noncoding RNAs: A Case Study of Unstable Angina Pectoris

Contact Info

Product

Resources

About