Prediction And Functional Analysis Of Prokaryotes Lysine Acetylation Based On Elastic Net

Lysine acetylation modification is a widely research and reversible protein post-translation modification(PTM),and has important significance in transcription regulation,cells metabolism,protein synthesis,cell cycle and cell morphology,signal transduction and many other physiological processes.The recognition of acetylation modification sites is the basis of understanding the molecular mechanism of acetylation proteins.Although more and more researchers have developed computational methods to identify the lysine acetylation sites in eukaryotes,few studies have been conducted on the prediction of lysine acetylation sites in prokaryotes.Based on the motif analysis of the data of lysine acetylation in prokaryotes and eukaryotes,we found that there were significant differences between the basal sites of lysine acetylation in prokaryotes and eukaryotes,which indicates that it is necessary to develop a reliable and efficient computational method for the prediction of lysine acetylation in prokaryotes.In this paper,based on the first order structure of lysine acetylated protein in prokaryotes,the prediction model was optimized by using Elastic net algorithm.Specific job content includes:1.Feature selection plays an important role in improving the performance of predictive models.This chapter mainly discusses the relationship between lasso algorithm,ridge regression algorithm and elastic net algorithm,and expounds the excellent properties of elastic network algorithm as a feature selection method.2.A new prediction lysine acetylation sites tool in different prokaryotes,named as ProAcePred(http://computbiol.ncu.edu.cn/ProAcePred),was developed based on elastic net algorithm.We collected the nine species experimental data for prokaryotes lysine acetylation from relevant literature and protein databases,using Elastic net algorithm to optimize the protein sequence characteristics,physicochemical characteristics and evolutionary information characteristics of lysine acetylated proteins in prokaryotes,then combined with support vector machine(SVM)classifier to construct the prediction model.The feature vector optimization by elastic network algorithm can improve the prediction performance of the model,and the feature analysis shows that evolutionary information plays an important role in the acetylation prediction model of prokaryotes,comparison with other existing methods shows that our prediction method has obvious advantages.3.Position specificity analysis method??information gain was used to improve the predictive quality of the acetylation sites of lysine in prokaryotes.We adopted a kind of effective position specificity analysis method??information gain(IG)algorithm,to optimize the acetylated peptides,and build a position specific window.Through combining the six different types feature,using the elastic network algorithm to optimize feature vector of model,we designed the prokaryotes lysine acetylation site prediction methods ProAcePred 2.0(http://computbiol.ncu.edu.cn/PAPred).The results of model training showed that the position-specific window was superior to the traditional continuous window and could effectively improve the predictive quality of the acetylation sites of the lysine in prokaryotes.