Computational Prediction of Thermodynamic Properties of Organic Molecules in Aqueous Solutions [Elektronische Ressource] / Ekaterina Ratkova. Gutachter: Eckhard Spohr ; Philippe A. Bopp. Betreuer: Maxim Fedorov

Frolov

Informations

• Publié par: universitat_duisburg-essen
• Publié le: 01 janvier 2011
• Nombre de lectures: 18
• Poids de l'ouvrage: 2 Mo

Extrait

ComputationalPredictionofThermodynamic
PropertiesofOrganicMoleculesinAqueous
Solutions
Dissertation
zurErlangungdesakademischenGradeseines
DoktorsderNaturwissenschaften
–Dr. rer. nat. –
vorgelegt von
EkaterinaL.Ratkova
geboreninZhdanov,USSR
FakultätfürChemie
der
UniversitätDuisburg-Essen
2011Die vorliegende Arbeit wurde im Zeitraum von Januar 2009 bis Mai 2011 im Arbeitskreis
von PhD, DSc, Priv.-Doz. Maxim V. Fedorov am Max-Planck-Institut für Mathematik in den
Naturwissenschaften,Leipzig,durchgeführt.
TagderDisputation: 21.07.2011
Gutachter: PhD,DSc,Priv.-Doz. MaximV.Fedorov
Prof. Dr. EckhardSpohr
Prof. Dr. Philippe A. Bopp
Vorsitzender: Prof. Dr. Eckart Hasselbrink CONTENTS 2
Contents
1 Introduction 7
2 TheoreticalBackground 16
2.1 MolecularOrnstein-Zernikeintegralequation.................. 16
2.2 3DReferenceInteractionSiteModel(3DRISM)................ 17
2.3 1DReferenceInteractionSiteModel(1DRISM) 19
2.4 HydrationFreeEnergyExpressionswithinthe1DRISMapproach ....... 22
2.5 Thermodynamicparameterswithinthe3DRISMapproach . . . . 25
2.6 PartialmolarvolumeexpressionsinRISMapproaches . . . . . ........ 25
3 ComputationalDetails 26
3.1 1DRISMcalculations............................... 26
3.2 3DRISMcalculations 28
3.3 Multigridtechnique................................ 29
3.3.1 One-levelPicarditerations . ....................... 31
3.3.2 Two-griditeration . ........................... 32
3.3.3 Multi-griditerations . 33
4 StructuralDescriptorsCorrection(SDC)model 35
4.1 TheQSPRbasisofthemodel . 35
4.2 Multi-parameterlinearregression......................... 39
4.3 Trainingandtestsets............................... 40
4.4 Choiceofdescriptors . 41
4.4.1 n−Alkanes................................. 41
4.4.2 Nonlinearalkanes. . ........................... 41
4.4.3 Othercompounds. . . 44
4.5 Optimalsetofcorrections . . . 47
5 ResultsandDiscussion 48
5.1 PMVestimationswith1Dand3DRISMapproaches .............. 48
5.2 1DRISM-SDCmodelwithOPLSpartialcharges................ 52
5.2.1 Calibrationofthemodel . . ....................... 52
5.2.2 Themodelpredictiveability . 55
5.2.3 ComparisonwithotherHFEexpressions . ............... 59CONTENTS 3
5.3 1DRISM-SDCmodelwithQM-derivedpartialcharges . . . . . . . . . . . . . 65
5.3.1 Performanceofthemodel . ....................... 65
5.3.2 Themodelpredictiveabilityforpollutants . ............... 72
5.4 3DRISM-SDCmodel............................... 81
5.4.1 Comparisonofuncorrecteddata . . ................... 81
5.4.2 Correctionforthecavityformation(3DRISM-UCmodel) ....... 82
5.4.3 Correctionsforthefunctionalgroups 86
5.4.4 The3DRISM-SDCmodelpredictiveability .............. 87
5.5 Comparisonofthemodelwiththecheminformaticsapproach . . . 92
6 Summary 94
7 Literature 99
8 Appendix1 119
9 Appendix2 150
9.1 ListofAbbreviations ...............................150
9.2 Shortsummary ..................................153
9.3 ListofPublications................................154
9.4 CurriculumVitae(CV)159
9.5 Erkärung . . .162
9.6 Acknowledgements163LISTOFFIGURES 4
ListofFigures
1 Thermodynamiccycleofadissolutionprocess. . . ............... 8
2 Hydrationfreeenergyinenvironmentalchemistry . . 9
3 Estimationsofexperimentalexpenses ...................... 12
4 Classificationofcomputationalmethods . . ................... 13
5 Correlationfunctionsinthe3Dand1DRISMapproaches . . . . ........ 18
6 SchemeofHFEcalculationswithintheRISMapproach . . . . . . 21
7 ParametersofRISMcalculations......................... 26
8 3DRISM.Benchmarkforthebufferdistance.................. 30
9 3DRISM.Benchmarkforthespacing...................... 30
10 QuantitativeStructure–PropertyRelationshipapproach . . . . ........ 36
11 SchematicrepresentationofmoleculeintheSDCmodel............ 38
12 Errorsconnectedwithnon-polarinteractions ................... 43
13 Errorswithspecificinteractions.................... 44
14 Setofstructuraldescriptors............................ 46
15 ContributionsofthePMV . ........................... 48
16 ComparisonofthecalculatedPMVsandexperimentaldata . . . ........ 51
17 ExamplesofSDCmodel’scontributions..................... 53
18 HFEscalculatedbythe1DRISM-SDC(OPLSq)modelforthetrainingset . . . 54
19 Partialchargesforheavyatomsofbenzene,phenol,andbenzylalcohol . . . . . 56
20 HFEspredictedbythe1Dmodelforthetestset . . . . . . 58
21 ComparisonofdifferentHFEexpressions . ................... 59
22 HFEscalculatedbythe1DRISM-SDC(GF)model............... 63
23 HFEsobtainedwiththeHNCclosure...................... 64
24 Representationoferrorsforphenolfragment . 67
25 1DRISM-SDCmodelwithdifferentcharges................... 68
26 Errorsofthemodelwithdifferentpartialcharges................ 68
27 Themodelpredictiveabilityforpolychlorinatedbenzenes . . . . ........ 74
28 ApparatusfortheHenry’slawconstantsdetermination . . . . . . 75
29 ExperimentaldataforPCBcongeners ...................... 78
30 HFEsofPCBsobtainedwithdifferentimplicitsolvationmodels........ 79
31 UncorrectedHFEscalculatedwithin1Dand3DRISMapproaches ....... 81
32 CorrelationbetweentheDPMVandtheHFEerrorforthe3DRISM . . . . . . 82LISTOFFIGURES 5
33 Thecorrelationfordruglikemolecules ...................... 83
34 ErrorsofHFEscalculatedbythecavitycorrectedHFEexpressions....... 86
35 Performanceofthe3DRISM-SDCmodelfortrainingset. . . . . 88
36 3DRISM-SDCmodelpredictiveability . . ................... 89
37 Errorscorrespondedtospecificinteractions. RISM-SDC(KH)models . . . . . 90
38 HFEscalculatedbytheRISM-SDC(KH)modelsfortrainingset........ 91
39 ComparisonoftheRISM-SDCmodelswiththecheminformaticsapproach. . . 93LISTOFTABLES 6
ListofTables
1 Correlationcoefficientsfordifferentsetsofdescriptors............. 47
2 ExperimentalvaluesofPMV........................... 49
3 Parametersofthe1DRISM-SDC(PW)model.................. 52
4 Statisticalprofileofthe1DRISM-SDC(OPLSq)model............. 57
5 Parametersofthe1DRISM-SDC(GF)model 60
6 Parametersofthe1DRISM-SDC(OPLSq)modelwiththeHNCclosure . . . . 62
7 Parametersof1DRISMcalculations....................... 62
8 Parametersofthe1DRISM-SDC(QMq)model................. 66
9 Comparisonofpartialchargesfor2-methylpropane ............... 69
10 OPLSandCHELPGpartialchargesfortoluene 70
11 Comparisonofpartialchargesforheterocyclicsolutes 71
12 1DRISM-SDC(QMq)modelparametersforpollutants............. 73
13 EfficiencyofdifferentimplicitmodelsforPCBs................. 76
14 ExperimentalandcalculatedHFEsforpolychlorinatedbenzenes........ 80
15 3DRISM-UC.ExperimentalandcalculatedHFEsfordruglikemolecules . . . 84
16 CoefficientsofthecavitycorrectedHFEexpression ............... 85
17 Parametersof3DRISM-SDCmodelsforspecificinteractions......... 87
18 3DRISM-SDCmodelwithKHHFEexpression . . . 92
19 Comparisonofthe1DRISM-SDCmodelthecheminformaticsapproach . . . . 94
20 Compositionofthetrainingsetandtestsetforthe1DRISM-SDCmodel . . . 119
21 Dataobtainedbythe1Dand3DRISM-SDCmodelforthetrainingset . . . . . 1311 INTRODUCTION 7
1 Introduction
Wateristhemostwidespreadandimportantmediaintheworld. Almostallglobalenvironmen-
talprocessesdealwithwater. Indeed,oceanscoverabout71%oftheEarthglobesurface,water
accumulatesintheskyformingclouds,anditaccessesalllandsontheEarthviaprecipitations.
Moreover, all biochemical processes take place in aqueous media: protein-ligand binding, par-
ticles transport in the blood stream, synthesis of biopolymers, etc. In chemical industry water
remainsoneofthemostwidelyusedsolvents[1].
The hydration free energy (HFE) is one of the key parameters characterizing the aqueous
solutionofasolute. First,HFEshowsthestrengthofsolute-waterinteractionswhichisimpor-
tant for such processes as biopolymer stabilization in aqueous solutions (proteins, DNA, etc.)
[2, 3, 4, 5, 6]. Second, HFE is crucial for the complex formation and binding processes taking
placeinaqueousmedia. Itdeterminesthefreeenergylossintheprocessofpartialdehydration
of interacting molecules which inevitably occurs during direct contact formation in solution
(e.g., ligand binding to a protein) [7, 8, 9]. Third, HFE of a compound determines partition of
the compound between gaseous and aqueous phases, and, thus, is significant for modeling of
molecules’ pathways in the environment (see the paragraph HFEinenvironmentalchemistry)
[10,11].
HFE equals the change of the Gibbs free energy that accompanies the transfer of solute
fromgaseousphasetoaqueoussolution[12]. Wenote,thattheamountofthetransferredsolute
molecules should be consistent with HFE units (e.g. HFE expressed in the terms of kcal/mol
correspondstothetransferof1moleofthesolutesmolecules).
HFEalsocanbedefinedfromthethermodynamiccycle: crystal–gaseousphase–solution
(Fig. 1). In this case, HFE can be derived in terms of two other thermodynamic properties:
sublimation free energy and solution free energy. Sublimation free energy (ΔG ) equals tosub
the change of the Gibbs free energy that accompanies the transfer of the solute from crystal to
gaseousphase,whilesolutionfreeenergy(ΔG )equalstothechangeoftheGibbsfreeenergysoln
thataccompaniesthetransferofthesolutefromcrystaltodilutedaqueoussolution(Fig. 1).
ΔG =ΔG −ΔG (1)hyd soln sub
Anotherimportantphysical