collectthisinformationrelatedwithsignalpathwaytoundersmndthecomprehensivesignalpathway.
Inthisdissertation,someoriginalresearchworkbytheauthor
asCanbeformulatedfclllow:
1.Oeneexpressionisonetheresultofsignalpathway,andtranscriptionfactors
onplaypivotrolesinthisprogress.AliteratureminingmethodbasedBayesianis
firstproposedtoretrievearticlesdescribingbindingsitesoftranscriptionfactor.By
statisticallyidentifyingwordsthatdiscriminate
eachnewabstractcanrelevantabstractsfromotherabstracts,scorethenbeassignedaloglikelihoodfordiscussingbindingsiteoftranscriptionfactors,Wejustattestthatthismethodis
onsimilarwithclassicalinformationretrievalmethodwhichisbasedTF/IDFtheory.Theefficiencyofthis
methodisimprovedgreatlybycombiningwithrelatedarticlesmethodof
TherecallrateandprecisionrateofOurmethod
traditionalkey-wordarePubMed.91%and45%,whichoutperforms
PubMedmethod(recallarate<=83%andprecisionrate<=26%respective).Althoughourmethodhaslittlelowerrecallratethansolerelatedarticlesof
.m-
(recallrate93%andprecisionrate27%)witll-2%,theprecisionrateishighwith-18%enhanced.We
2.Proteinfoundabout63.000interestedarticleswiththismethod.importantroleintransforminginformationofkinases(PKs)play
signalpathway.Theyphosphorylatethesubstrates(proteins)atthespecificsites(phosphorylationsites)flankingwithcanonicalmotif.Onceagain,Weminedliterature
USedescribingphosphorylatesitesbyusingtheBayesianmethod.Wealsobuiltall
assistanttooltoaddcolortasinsentencesforrapid
dataanddatainprocess.WithonthesearchresultPhospho.ELM,WeproposedamethodbasedBayesiandecisiontheoIy_PPSPtopredictthepotentialphosphorylationsitesofPKs。Predictionresultson~70PKgroupsshowthatingeneral,itoutperformsstateoftheartmethods:Scansite,KinasePhos,NetPhosKandGPS,whichsuggeststhatthismethodisanothercompetitivecomputationalapproachinthisbranchofbioinformaties.Atthesametime,thismethodhastheadvantageofsimpleness,efficiencyandrobustness.Awebserviceisalsoavailableforonlineperditionat(http://bioinformaties.1cd-ustc0rg/旦鳖).
3.Anovelmethodcalled“TranscriptionFactor-MediatedPathwayAnalysis’’ispresentedtoinferabnormaltranscriptionfactorsandpathwaysincancerchips.Theactivityofatranscriptionfactorisinferredbyevaluatingthenetresultofpercentsof
genesinaactivated(orrepressed)targetchip,andthentheabnormaltranscription
thefactorismappedtopathwaysdepositedinKEGGtThisalgorithmintegrates
experimentsofgene—regulationand
breastpathway.Wecancerhaveanalyzedhumangastriccancer,MicroarraycancerandIIdifferenttypesofwhichstoredinStanford
Database(sMD)in
pathwayarethismethodandfoundthatTGF-8,JAK-STAT,NF-?BandNotchover-activatedinmanyofthesechips.Theseabnormalpathwayswillbeofgreathelpinunderstandingtheprogressofcancerandinrationaldmgdesign.