预览加载中,请您耐心等待几秒...
1/2
2/2

在线预览结束,喜欢就下载吧,查找使用更方便

如果您无法下载资料,请参考说明:

1、部分资料下载需要金币,请确保您的账户上有足够的金币

2、已购买过的文档,再次下载不重复扣费

3、资料包下载后请先用软件解压,在使用对应软件打开

联合特征驱动方法和模板方法预测蛋白质的核酸绑定残基 Introduction: Protein-nucleicacidinteractionsplayacrucialroleinmanybiologicalprocesses,suchasDNAreplication,transcription,andtranslation.Identificationandpredictionofnucleicacid-bindingresiduesinproteinscanprovideimportantinsightsinunderstandingthesebiochemicalprocesses.Machinelearning-basedapproacheshavebeenwidelyusedinpredictingnucleicacid-bindingresiduesinproteins.Inthiswork,weproposeaunifiedmethodcombiningfeature-drivenapproachandtemplate-basedapproachtoimprovethepredictionaccuracyofnucleicacid-bindingresiduesinproteins. Feature-DrivenApproach: Infeature-drivenapproach,weextractdifferentstructuralandphysicochemicalfeaturesofaminoacidstopredictwhetheragivenaminoacidresidueisanucleicacid-bindingresidueornot.Thefeaturesweusedinthisstudyincludeaminoacidconservationscore,solventaccessibility,secondarystructure,hydrogenbonding,andelectrostaticpotential.Thesefeaturesareknowntoplayacrucialroleinnucleicacidbinding. Weusedtherandomforestclassifiertotrainthemodelusingtheextractedfeatures.Themodelachievedanaccuracyof75%inidentifyingnucleicacid-bindingresiduesinproteins.However,theaccuracyofthemodelwasstillnotsatisfactory. Template-BasedApproach: Intemplate-basedapproach,weusedknownnucleicacid-bindingproteinstructuresastemplatestopredictnucleicacid-bindingresiduesinthetargetproteinsequence.WefirstlyidentifiedstructuraltemplatesfromthePDBlibrarythataresimilartothetargetsequence.Then,wemappedthetemplatestothetargetsequenceandpredictednucleicacid-bindingresiduesbasedontheobservednucleicacid-bindingresiduesinthetemplates. Thetemplate-basedapproachachievedanaccuracyof85%inpredictingnucleicacid-bindingresidues.However,thedrawbackofthisapproachisthatitheavilyreliesontheavailabilityofstructuraltemplatesinthePDBlibrary. UnifiedMethod: Weproposedaunifiedmethodthatcombinesthefeature-drivenapproachandthetemplate-basedapproachtoimprovethepredictionaccuracyofnucleicacid-bindingresiduesinproteins.TheunifiedmethodfirstidentifiesstructuraltemplatesfromthePDBlibrarya