Predicting essential proteins based on subcellular localization, orthology and PPI networks

Li, Gaoshi; Li, Min; Wang, Jianxin; Wu, Jingli; Wu, Fang-Xiang; Pan, Yi

doi:10.1186/s12859-016-1115-5

Cited by 75 publications

(51 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…gives the results. From table4 we can see that DeepHE (N+S) significantly outperforms the other machine learning models regarding to three comprehensive measures, AUC, AP, and Accuracy. For instance, the AUC score of DeepHE (N+S) is 8.79% higher than that of SVM (N+S), 43.22% higher than that of NB (N+S), 25.38% higher than that of RF (N+S), and 15.39% higher than that of Adaboost (N+S).…”

mentioning

confidence: 95%

DeepHE: Accurately Predicting Human Essential Genes based on Deep Learning

Zhang

Xiao

2020

Preprint

View full text Add to dashboard Cite

Motivation: Accurately predicting essential genes using computational methods can greatly reduce the effort in finding them via wet experiments at both time and resource scales, and further accelerate the process of drug discovery. Several computational methods have been proposed for predicting essential genes in model organisms by integrating multiple biological data sources either via centrality measures or machine learning based methods. However, the methods aiming to predict human essential genes are still limited and the performance still need improve. In addition, most of the machine learning based essential gene prediction methods are lack of skills to handle the imbalanced learning issue inherent in the essential gene prediction problem, which might be one factor affecting their performance. Results:We proposed a deep learning based method, DeepHE, to predict human essential genes by integrating features derived from sequence data and protein-protein interaction (PPI) network. A deep learning based network embedding method was utilized to automatically learn features from PPI network. In addition, 89 sequence features were derived from DNA sequence and protein sequence for each gene. These two types of features were integrated to train a multilayer neural network. A cost-sensitive technique was used to address the imbalanced learning problem when training the deep neural network. The experimental results for predicting human essential genes showed that our proposed method, DeepHE, can accurately predict human gene essentiality with an average AUC higher than 94%, the area under precision-recall curve (AP) higher than 90%, and the accuracy higher than 90%. We also compared DeepHE with several widely used traditional machine learning models (SVM, Naïve Bayes, Random Forest, Adaboost). The experimental results showed that DeepHE greatly outperformed the compared machine learning models. Conclusions:We demonstrated that human essential genes can be accurately predicted by designing effective machine learning algorithm and integrating representative features captured from available biological data. The proposed deep learning framework is effective for such task.Essential genes are a subset of genes which are indispensable to the survival or reproduction of a living organism. The prediction of gene essentiality is very important for understanding the minimal requirements of an organism, identifying disease genes, and finding new drug targets. The discovery of essential genes via wet-lab experimental methods are often time-consuming, laborious, and costly. With the accumulation of gene essentiality data in some model organisms and human cell lines, many computational methods have been proposed to predict essential genes by exploring the correlations between gene essentiality and all sorts of biological information.One focus in this direction is network based centrality measures. Many studies have demonstrated that highly connected proteins in a protein-protein interaction (PPI) network are more likely to be esse...

show abstract

mentioning

confidence: 95%

DeepHE: Accurately Predicting Human Essential Genes based on Deep Learning

Zhang

Xiao

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…CoTB exhibited the best performance and obtained prediction precisions of 89, 78, 79, and 85 percent on the YDIP, YMIPS, YMBD and YHQ datasets at the top 100 level, respectively. In particular, compared to the most recently developed method, SON [ 23 ], CoTB improved the prediction precisions by at least 9, 10, 8, 8, 7, and 8 percent on the YDIP dataset at the top 100 to top 600 levels, respectively. Compared to GOS [ 24 ], CoTB improved the prediction precisions by at least 6, 6, 9, and 10 percent on the YDIP dataset at the top 300 to top 600 levels, respectively.…”

Section: Introductionmentioning

confidence: 96%

“…It has been proven that orthologous properties are positively correlated with protein essentiality [ 22 ]. Li [ 23 ] proposed a method named SON that integrates subcellular localization and orthologous score (OS) information, and this method improved the accuracy of predicting essential proteins to approximately 81 percent on the YDIP dataset at the top 100 level. Li [ 24 ] proposed a method named GOS that integrates gene expression, orthology, and subcellular localization information to identify essential proteins.…”

Section: Introductionmentioning

confidence: 99%

A new computational strategy for identifying essential proteins based on network topological properties and biological information

2017

View full text Add to dashboard Cite

Essential proteins are the proteins that are indispensable to the survival and development of an organism. Deleting a single essential protein will cause lethality or infertility. Identifying and analysing essential proteins are key to understanding the molecular mechanisms of living cells. There are two types of methods for predicting essential proteins: experimental methods, which require considerable time and resources, and computational methods, which overcome the shortcomings of experimental methods. However, the prediction accuracy of computational methods for essential proteins requires further improvement. In this paper, we propose a new computational strategy named CoTB for identifying essential proteins based on a combination of topological properties, subcellular localization information and orthologous protein information. First, we introduce several topological properties of the protein-protein interaction (PPI) network. Second, we propose new methods for measuring orthologous information and subcellular localization and a new computational strategy that uses a random forest prediction model to obtain a probability score for the proteins being essential. Finally, we conduct experiments on four different Saccharomyces cerevisiae datasets. The experimental results demonstrate that our strategy for identifying essential proteins outperforms traditional computational methods and the most recently developed method, SON. In particular, our strategy improves the prediction accuracy to 89, 78, 79, and 85 percent on the YDIP, YMIPS, YMBD and YHQ datasets at the top 100 level, respectively.

show abstract

“…The subcellular localization diversity of the protein products of a gene may serve as an important characteristic in revealing gene essentiality, in the context of subcellular locations. Previous studies have discussed the protein localization specificity in prokaryotes (Peng and Gao, 2014), and integrated the subcellular localized information with PPI topology for predicting essential genes (Acencio and Lemke, 2009;Li et al, 2016;Li et al, 2018), using annotation data from the Gene Ontology-Cellular Component Ontology (GO-CCO) (Ashburner et al, 2000). Although the priority of these components definitely needs to be discussed, another potentially important characteristic of protein localization seems to be ignored, which is the subcellular diversity.…”

Section: Introductionmentioning

confidence: 99%

Quantifying Gene Essentiality Based on the Context of Cellular Components

2020

View full text Add to dashboard Cite

Different genes have their protein products localized in various subcellular compartments. The diversity in protein localization may serve as a gene characteristic, revealing gene essentiality from a subcellular perspective. To measure this diversity, we introduced a Subcellular Diversity Index (SDI) based on the Gene Ontology-Cellular Component Ontology (GO-CCO) and a semantic similarity measure of GO terms. Analyses revealed that SDI of human genes was well correlated with some known measures of gene essentiality, including protein-protein interaction (PPI) network topology measurements, dN/dS ratio, homologous gene number, expression level and tissue specificity. In addition, SDI had a good performance in predicting human essential genes (AUC = 0.702) and drug target genes (AUC = 0.704), and drug targets with higher SDI scores tended to cause more side-effects. The results suggest that SDI could be used to identify novel drug targets and to guide the filtering of drug targets with fewer potential side effects. Finally, we developed a user-friendly online database for querying SDI score for genes across eight species, and the predicted probabilities of human drug target based on SDI.

show abstract

Predicting essential proteins based on subcellular localization, orthology and PPI networks

Cited by 75 publications

References 57 publications

DeepHE: Accurately Predicting Human Essential Genes based on Deep Learning

DeepHE: Accurately Predicting Human Essential Genes based on Deep Learning

A new computational strategy for identifying essential proteins based on network topological properties and biological information

Quantifying Gene Essentiality Based on the Context of Cellular Components

Contact Info

Product

Resources

About