A High Efficient Biological Language Model for Predicting Protein–Protein Interactions

Wang, Yanbin; You, Zhu‐Hong; Yang, Shan; Xiao, Li; Jiang, Tongwen; Zhou, Xi

doi:10.3390/cells8020122

Cited by 75 publications

(58 citation statements)

References 31 publications

(28 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Networks are ubiquitous in the real world, and can be organized in the form of graphs where nodes represent various objects and edges represent relationships between objects. For examples, in a protein-protein interaction network (Wang et al, 2019), the physical interactions among proteins constitute the networks of protein complexes where each individual protein is an independent node and the interaction represents an edge. In medical practice (Litjens et al, 2017), analyzing protein-protein networks can gain new insights into biochemical cascades and guide the discovery of putative protein targets of therapeutic interest.…”

Section: Introductionmentioning

confidence: 99%

Multi-Task Network Representation Learning

et al. 2020

View full text Add to dashboard Cite

Networks, such as social networks, biochemical networks, and protein-protein interaction networks are ubiquitous in the real world. Network representation learning aims to embed nodes in a network as low-dimensional, dense, real-valued vectors, and facilitate downstream network analysis. The existing embedding methods commonly endeavor to capture structure information in a network, but lack of consideration of subsequent tasks and synergies between these tasks, which are of equal importance for learning desirable network representations. To address this issue, we propose a novel multi-task network representation learning (MTNRL) framework, which is end-to-end and more effective for underlying tasks. The original network and the incomplete network share a unified embedding layer followed by node classification and link prediction tasks that simultaneously perform on the embedding vectors. By optimizing the multi-task loss function, our framework jointly learns task-oriented embedding representations for each node. Besides, our framework is suitable for all network embedding methods, and the experiment results on several benchmark datasets demonstrate the effectiveness of the proposed framework compared with state-of-the-art methods.

show abstract

Section: Introductionmentioning

confidence: 99%

Multi-Task Network Representation Learning

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Currently, many kinds of computational models based on protein sequences have been presented for predicting PPIs. In this section, to further objectively validate the prediction performance of the proposed method, seven state-of-the-art methods, including Ensemble Deep Neural Networks (EnsDNN) [22], 3-mers-based [31], Bio2vec-based [31], pseudo Substitution Matrix Representation (pseudo-SMR) [32], WSRC with continuous wavelet and discrete wavelet transform (WSRC+CW and DW) [33], feature weighted rotation forest algorithm (FWRF) [17], and Global encoding [34] were compared on the human, H. pylori, and yeast data sets. The comparison results of three benchmark data sets based on five-fold cross-validation of different models are plotted in Figures 2-4, respectively.…”

Section: Comparison With Other Methodsmentioning

confidence: 99%

Protein-Protein Interactions Prediction Based on Graph Energy and Protein Sequence Information

Xu¹,

Xu²,

Zhang³

et al. 2020

Molecules

View full text Add to dashboard Cite

Identification of protein-protein interactions (PPIs) plays an essential role in the understanding of protein functions and cellular biological activities. However, the traditional experiment-based methods are time-consuming and laborious. Therefore, developing new reliable computational approaches has great practical significance for the identification of PPIs. In this paper, a novel prediction method is proposed for predicting PPIs using graph energy, named PPI-GE. Particularly, in the process of feature extraction, we designed two new feature extraction methods, the physicochemical graph energy based on the ionization equilibrium constant and isoelectric point and the contact graph energy based on the contact information of amino acids. The dipeptide composition method was used for order information of amino acids. After multi-information fusion, principal component analysis (PCA) was implemented for eliminating noise and a robust weighted sparse representation-based classification (WSRC) classifier was applied for sample classification. The prediction accuracies based on the five-fold cross-validation of the human, Helicobacter pylori (H. pylori), and yeast data sets were 99.49%, 97.15%, and 99.56%, respectively. In addition, in five independent data sets and two significant PPI networks, the comparative experimental results also demonstrate that PPI-GE obtained better performance than the compared methods.

show abstract

“…Meanwhile, we applied a wide range of evaluation criteria to effectively assess the performance of our method (Wang, et al, 2019). Cross validation is a widely used method to measure model ability You, et al, 2017).…”

Section: Evaluation Criteriamentioning

confidence: 99%

MeSHHeading2vec: A new method for representing MeSH headings as feature vectors based on graph embedding algorithm

Guo

You

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

Motivation: Effectively representing the MeSH headings (terms) such as disease and drug as discriminative vectors could greatly improve the performance of downstream computational prediction models. However, these terms are often abstract and difficult to quantify. Results: In this paper, we converted the MeSH tree structure into a relationship network and applied several graph embedding algorithms on it to represent these terms. Specifically, the relationship network consisting of nodes (MeSH headings) and edges (relationships) which can be constructed by the rule of tree num. Then, five graph embedding algorithms including DeepWalk (DW), LINE, SDNE, LAP and HOPE were implemented on the relationship network to represent MeSH headings as vectors. In order to evaluate the performance of the proposed method, we carried out the node classification and relationship prediction tasks. The experimental results show that the MeSH headings characterized by graph embedding algorithms can not only be treated as an independent carrier for representation, but also can be utilized as additional information to enhance the distinguishable ability of vectors. Thus, it can act as input and continue to play a significant role in any disease-, drug-, microbe-and etc.-related computational models. Besides, our method holds great hope to inspire relevant researchers to study the representation of terms in this network perspective. Contact: zhuhongyou@ms.xjb.ac.cn

show abstract

A High Efficient Biological Language Model for Predicting Protein–Protein Interactions

Cited by 75 publications

References 31 publications

Multi-Task Network Representation Learning

Multi-Task Network Representation Learning

Protein-Protein Interactions Prediction Based on Graph Energy and Protein Sequence Information

MeSHHeading2vec: A new method for representing MeSH headings as feature vectors based on graph embedding algorithm

Contact Info

Product

Resources

About