Research On ?-Lactamase Prediction And Annotation Analysis Based On Ensemble Learning

?-lactamase is a kind of protease secreted by pathogenic bacteria.It increases bacterial resistance by binding to the lactam ring of the ?-lactam class of antibiotics,which destroys antibiotic activity.The continuous mutation and evolution of ?-lactamases has caused an increasing number of antibiotic drugs to fail,which has attracted widespread attention from researchers.In recent years,there are many prediction tools for identifying potential ?-lactamase proteins,but the performance of them are not satisfactory due to their straightforward models.Besides,there is no platform trying to combine not only the prediction but also the downstream analysis,to further annotate and analysis the predicted?-lactamases.For the above reasons,this paper proposed a multi-classification method based on ensemble learning,which can identify the ?-lactamase according to the protein sequence,and the prediction result is accurate to its sub-category.At the same time,in this research we constructed BLHub(Beta-Lactamase Hub),a comprehensive platform with the function of protein annotation,prediction and specific analysis,in order to realize the integrated analysis of ?-lactamase.The main research contents and conclusions of this paper are as follows:(1)Prediction for ?-lactamase.This paper proposes an integrated learning model that can predict ?-lactamase proteins and further classify the subtype of ?-lactamase proteins.The model extracts features from the sequence,physical and chemical properties and its evolutionary information.Then,four traditional machine learning methods including random forest,support vector machine,naive Bayes and K nearest neighbors are integrated through stacking ensemble strategy.The results of five-fold cross-validation and independent test show that the whole performance of the stacking ensemble model is better than that of a single machine learning model.(2)Construction of the ?-lactamase analysis platform.This paper builds a ?-lactamase protein analysis platform BLHub(http://47.110.242.168/index.jsp).The platform,is implemented with the Java language,uses Bootstrap 3.3.6,Ajax,j Query,My SQL and Strust2 technologies to realize the interaction between function and data of the front end and back end.The platform includes rich and multi-sourceprotein annotation information,such as protein structure information,protein active sites,protein classification genealogy,and antibiotic resistance information.Then,the platform embeds a stacking ensemble predictor so as to discover potential ?-lactamase proteins and identify their subtype.In addition,in order to facilitate scientific researchers to further infer the function and structure of the unknown protein,the platform provides analysis methods based on protein sequence including similarity analysis and affinity analysis.It is hoped that the prediction method proposed in this paper and the comprehensive analysis platform constructed will be a useful tool for researchers to study ?-lactamase protein and further accelerate the discovery and exploration of potential ?-lactamase protein.