Protein Function Prediction Using Function Associations in Protein–Protein Interaction Network

Sun, Ping; Xian, Tan; Guo, Sijia; Zhang, Jingbo; Sun, Bing; Du, Ning; Wang, Han; Sun, Hui

doi:10.1109/access.2018.2806478

Cited by 13 publications

(6 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Computational methods are also applied due to low cost, flexibility, easy implementation and large datasets [16], [17]. Recent work in [18] predicts the protein function based on an iterative algorithm. Ma et al [19] employed support vector machine based classifiers to correlate the energy binding features with geometrical shape of the protein.…”

Section: Prior Workmentioning

confidence: 99%

Parametric Models for Understanding Atomic Trajectories in Different Domains of Lung Cancer Causing Protein

et al. 2019

View full text Add to dashboard Cite

Non-small cell lung cancer (NSCLC) is a major cause of death worldwide. About 80% to 85% of lung cancer cases are NSCLC. It is well known that mutation of the epidermal growth factor (EGFR) may lead to the NSCLC. The first generation drugs are effective initially, but almost all patients develop drug resistance after about a year due to a secondary mutation. The computational methods are an efficient tool for investigating drug resistance, design, and discovery. Moreover, molecular dynamics (MD) simulation enables us to study and analyze the behavior of proteins and molecules at the atomic level. MD simulations offer extraordinary insight about biomolecules and are a valuable tool for computer aided drug discovery. Earlier studies on EGFR only focused on the kinase domain. Because EGFR is a multi-domain protein, mutations in the kinase domain may affect the function in other domains. Therefore, it is important to investigate the complete structure of the EGFR and its mutants. In this paper, we first generate the complete structure of the EGFR and perform MD simulation for the wildtype EGFR, EGFR with L858R mutation and EGFR with L858R and T790M mutation. We divide the complete structure of the EGFR and its mutants into 8 domains according to the reference crystal structure. We then consider atom trajectories as time series signals and estimate the power spectral densities using the auto-regressive integrated (ARI) model, which shows interesting insight. Dynamic time warping is used to analyze the similarity between each domain of the structures. Interesting patterns are observed which may be useful for investigating drug resistance and design. Furthermore, Pearson correlation coefficient, peaks, and widths of the power spectral density are calculated for each domain. The simulation results provide useful insight about conformation dynamics of the EGFR, such as atom motion and protein stability. The domains are less correlated in L858R type and even weaker when the second mutation occurs. The warping patterns are changed due to mutation and the movement of atoms is distorted. Hence, it is difficult for a drug to bind to the protein. These findings will be useful in understanding the characteristics of the EGFR and for computer aided drug design process for the NSCLC patients. INDEX TERMS Autoregressive integrated model, drug resistance, dynamic time warping, epidermal growth factor receptor, molecular dynamics, non-small cell lung cancer.

show abstract

Section: Prior Workmentioning

confidence: 99%

Parametric Models for Understanding Atomic Trajectories in Different Domains of Lung Cancer Causing Protein

et al. 2019

View full text Add to dashboard Cite

show abstract

“…This method operated in two phases: Spectral clustering was used to cluster the PPI network followed by the application of the betweenness centrality measure for labeling within each cluster, and then the labeled protein data were used by a classification algorithm. Associations between functions in a PPI network were used in [12], stating that multiple function labels assigned to proteins were not independent and their coexistence could be used effectively to predict protein function. A deep semantic text representation was presented in [13], with various pieces of information extracted from protein sequences such as homology, motifs, and domains.…”

Section: Introductionmentioning

confidence: 99%

A Neural Network-Based Multi-Label Classifier for Protein Function Prediction

Tahzeeb

Hasan

2022

Eng. Technol. Appl. Sci. Res.

View full text Add to dashboard Cite

Knowledge of the functions of proteins plays a vital role in gaining a deep insight into many biological studies. However, wet lab determination of protein function is prohibitively laborious, time-consuming, and costly. These challenges have created opportunities for automated prediction of protein functions, and many computational techniques have been explored. These techniques entail excessive computational resources and turnaround times. The current study compares the performance of various neural networks on predicting protein function. These networks were trained and tested on a large dataset of reviewed protein entries from nine bacterial phyla, obtained from the Universal Protein Resource Knowledgebase (UniProtKB). Each protein instance was associated with multiple terms of the molecular function of Gene Ontology (GO), making the problem a multilabel classification one. The results in this dataset showed the superior performance of single-layer neural networks having a modest number of neurons. Moreover, a useful set of features that can be deployed for efficient protein function prediction was discovered.

show abstract

“…In the last 20 years, many experimental methods have been developed to reveal the high‐quality structure of PPI networks in many organisms, such as humans and yeast [ 2 , 3 , 4 , 5 ]. The exponential growth in biotechnology has led to the availability of a wide range of databases describing PPI networks [ 6 , 7 ]. Therefore, system‐level representation of the PPI network provides an opportunity to select a subset of genes that play an important role in cell viability, such as essential genes and cancer target genes [ 8 ].…”

Section: Introductionmentioning

confidence: 99%

Constraint‐based models for dominating protein interaction networks

Alofairi

Mabrouk

Elsemman

2021

IET Systems Biology

View full text Add to dashboard Cite

The minimum dominating set (MDSet) comprises the smallest number of graph nodes, where other graph nodes are connected with at least one MDSet node. The MDSet has been successfully applied to extract proteins that control protein-protein interaction (PPI) networks and to reveal the correlation between structural analysis and biological functions. Although the PPI network contains many MDSets, the identification of multiple MDSets is an NP-complete problem, and it is difficult to determine the best MDSets, enriched with biological functions. Therefore, the MDSet model needs to be further expanded and validated to find constrained solutions that differ from those generated by the traditional models. Moreover, by identifying the critical set of the network, the set of nodes common to all MDSets can be time-consuming. Herein, the authors adopted the minimisation of metabolic adjustment (MOMA) algorithm to develop a new framework, called maximisation of interaction adjustment (MOIA). In MOIA, they provide three models; the first one generates two MDSets with a minimum number of shared proteins, the second model generates constrained multiple MDSets (k-MDSets), and the third model generates user-defined MDSets, containing the maximum number of essential genes and/or other important genes of the PPI network. In practice, these models significantly reduce the cost of finding the critical set and classifying the graph nodes. Herein, the authors termed the critical set as the k-critical set, where k is the number of MDSets generated by the proposed model. Then, they defined a new set of proteins called the ðk − 1Þ-critical set, where each node belongs to ðk − 1Þ MDSets. This set has been shown to be as important as the k-critical set and contains many essential genes, transcription factors, and protein kinases as the k-critical set. The ðk − 1Þ-critical set can be used to extend the search for drug target proteins. Based on the performance of the MOIA models, the authors believe the proposed methods contribute to answering key questions about the MDSets of PPI networks, and their results and analysis can be extended to other network types.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

show abstract

Protein Function Prediction Using Function Associations in Protein–Protein Interaction Network

Cited by 13 publications

References 20 publications

Parametric Models for Understanding Atomic Trajectories in Different Domains of Lung Cancer Causing Protein

Parametric Models for Understanding Atomic Trajectories in Different Domains of Lung Cancer Causing Protein

A Neural Network-Based Multi-Label Classifier for Protein Function Prediction

Constraint‐based models for dominating protein interaction networks

Contact Info

Product

Resources

About