Inthisworkamulti-objectivequantitativestructure-propertyrelationship(QSPR)analysisapproachwasreportedbasedonthestudyonthreepartitionpropertiesof50aromaticsulfur-containingcarboxylates.Heremulti-objectives(properties)weretakenasavectorforQSPRmodeling.Thequantitativecorrelationsforpartitionpropertiesweredevelopedusingage-neticalgorithm-basedvariable-selectionapproachwithquantumdescriptors,derivedfromAM1-basedcalculations.WiththeQSPRmodels,theaqueoussolubmty,octanol/waterpartitioncoefficientsandreversed-phaseHPLCcapacityfactorsofsulfur-contalningcompoundswereestimatedandpredicted.UsingGA-basedmultivariatelinearregressionwithcross-vali-dationprocedure,asetofthemostpromisingdescriptorswasselegtedfromapoolof28quantumchemicalsemi-empiricalde-scriptors,inclodingstericandelectronictypes,tointegrallybuildQSPRmodels.Theselectedmoleculardescriptorsinclud-edthenetchargesoncarboxylgroup(Qoc),the2ndpowerofnetehnrgesonnitrogenatoms(QN^2),thenetatomicchargeonthesulfuratoms(Qs),thevanderWaalsvolumeofmolecule(V),themostpositivenetatomicchargeonhydrogenatoms(QH)andthemeasureofpolarityandpolarizability(π),whichweremainfactorsaffectingthedistributionprocessesofthecompoundsunderstudy.ThestatisticallybestQSPRmodelsofsixdescriptorsweresimultaneouslyobtainedbyGA-basedlinearregressionanalysis.WiththeselecteddescriptorsandtheQSPRequations,mechanismsofpartitionactionoftheSulfur-containingcarboxylateswereabletobeinvestigatedandinter-preted.
