摘要
Thestudyofmetalcompoundsandantibiotics(targetmolecules)asanti-cancerdrugwasameaningandremarkabletopics.TheinteractionofthesetargetmoleculesandDNAwasalwaysafoucsthatpeoplepaiedcloseattentionto.AnunderstandingofhowmoleculesbindtoDNAwouldbepotentiallyusefulinthedesignandfilteringnewdrugs,medicinesynthesizestherapyandgeneticdiseases,etc.Thebindingmode,functionmechanism,bindingsitepointandbindingconstantarethemaincontentsofthesekindsofstudy.Uptonow,fluorescencespectrum,ultraviolet-visiblespectroscopy,capillaryelectrophoresisandelectrochemicaltechniqueshavebeendevelopedfortheidentificationandcharacterizationofinteractionsbetweenthem,andhavemadegreatprogress.However,therearestillsomeproblemsinthisfield.First,itisdifficulttofindtheinfluencingfactorsfromthestructureofmolecules.Second,forbatchesofunknownmolecules,itisdifficulttopredicttheinteractionfromtheexperimentalresults.Inthispaper,principalcomponentsanalysis,multivariateanalysis,artificialneuralnetworkandhierarchicalclusteranalysiswereappliedtostudytheinteractionoftargetmoleculesandDNAandtheirinfluencingfactors.Themainworksofthispaper-wereasfollows:PartI:IntroductionThispartrelatedthedevelopinghistoryofchemicalpatternrecognitiontechniques(principalcomponentsanalysis,artificialnervesnetandhierarchicalclusteranalysis),introducedtheirbasicmathematicaltheoryandsummeduptheirapplicationinthefieldofanalyticalchemistry.PartII:PredictivestudiesoninteractionbetweenDNAandmetalcompoundsThestudyoftheinteractionofmetalcompoundsandDNAhasveryhightheoriesworthandpracticalvalue.ThewayandthemechanismoftheinteractionofDNAandmetalcompound,andtheirgeneralregulationwereveryusefultoinvestigatetheapplicationofcompoundinanti-tumormedicine,molecularbiology,techniqueandotherrelatedfieldofthelivingcreatureengineering.Inthispaper,63metalcompoundswereselectedasresearchobjecttodiscussmaininfluencingfactorsoftheirinteractionwithDNAfrommolecularstructure.Principalcomponentsanalysis(PCA)wasemployedinordertoreducedimensionalityandinvestigatewhichsubsetofvariablescouldbemoreeffectiveforclassifyingthecompoundaccordingtotheirbindingmodewithDNA.Artificialnervesnet(ANN)andmultivariateanalysis(MLR)wereappliedtoestablishtwopredictingconstantmodelsandonepredictingbindingmodemodel.21parameters(descriptors)thatcaninfluencetheinteractionofselectedmetalcompoundandDNAhavebeeninvestigated.ResultsofthePCAstudyshowedthatthereare10descriptorscloselyrelatedtotheinteraction.Theproposedmodelshadthegoodaccuracyinpredictingconstantandmode.PartIII:StudiesontheinteractionbetweenantibioticsandDNAInteractionbetweenantibioticmoleculesandDNAhastheimportantsignificanceonunderstandingdrugfunction,designingmolecularandfilteringdrugs.Thepresentstudiesaimedtofindinteractionmechanismbyexperimentaltechniques.Butitisdifficulttofindthestructureinfluencefactorsthataffecttheinteraction,andforbatchesoftargetmolecular,theinteractioninformationisdifficulttopredict.Inthisstudy,basedonexperimentaldataandquantifiedstructuredata,principalcomponentsanalysis,ANN,HCAandMLRwereappliedtostudytheinteraction.Theinfluencefactorswereinvestigated.Twomodelsofpredictinginteractionconstantandonemodelofpredictinginteractionmodewereestablished.Among24quantifiedparameters,itwasfoundthattherewere12parametersthataffectedtheinteractionmarkedly.Forthepredictionofbindingconstant,theminimalrelativeerrorisonly0.17%(MLR)and0.72%(ANN);forthepredictionofbindingmode(ANN),fivetestmoleculesareallconsistentwithexperimentalresults.PartIV:PredicttheinteractiveconstantofDNAandheterocyclicmoleculesbymodelAccordingtothepredictiveconstantmodelbypreviousstudies,30bindingconstantoftheinteractionofheterocyclicmolecules(dyecompounds)andDNAwerepredictedinthisstudy.Meanwhile,wediscussedtheeffectoftheinteractionoftargetmoleculebytheinfluencedfactorswhichhadbeenfound.Theinvestigationsinthisworkprovideworthyinformationtodesignandfiltermedicinemolecules.
-
单位西北师范大学