La lecture à portée de main
Description
Sujets
Informations
Publié par | universitat_potsdam |
Publié le | 01 janvier 2008 |
Nombre de lectures | 25 |
Poids de l'ouvrage | 3 Mo |
Extrait
ComparativeAnalysisofM olecular
InteractionNetworks:TheI nterplay
BetweenSpatialandFunctional
OrganizingPrinciples.
VergleichendeAnalysemolekularerInteraktionsnetzwerke:Der
ZusammenhangvonräumlichenundfunktionellenOrganisationsprinzipien.
Dissertation
zurErlangungdesGrades
DoktorderNaturwissenschaften(Dr.rer.nat)
eingereichtam
InstitutfürBiochemieundBiologieander
Mathematisch.NaturwissenschaftlichenFakultät
UniversitätPotsdam
vorgelegtvon
DIPL. ING. PAWEL DUREK
Die vorliegende Arbeit wurde angefertigt am
Max-Planck-Institut für Molekulare Pflanzenphysiologie
Potsdam, Dezember 2008
This work is licensed under a Creative Commons License:
Attribution - Noncommercial - No Derivative Works 3.0 Germany
To view a copy of this license visit
http://creativecommons.org/licenses/by-nc-nd/3.0/de/deed.en
Published online at the
Institutional Repository of the University of Potsdam:
http://opus.kobv.de/ubp/volltexte/2009/3143/
urn:nbn:de:kobv:517-opus-31439
[http://nbn-resolving.de/urn:nbn:de:kobv:517-opus-31439] Contents
Publications 1
Abstract 3
Zusammenfassung 5
Chapter1
Introduction 7
1.1ProteinInteractionNetworks 8
1.2MetabolicInteractionNetworks 10
1.3PhosphorylationNetworks 12
1.4ThesisOverview 15
Chapter2
GraphtheoreticalconceptsinthecontextofBiologicalNetworks 17
2.1Topologicalpropertiesofgraphs 18
2.2Centralityofnodes 20
Chapter3
The integrated analysis of metabolic and protein interaction networks
revealsnovelmolecularorganizingprinciples 23
3.1Background 24
3.2Results 26
TopologicalPropertiesofInteractionNetworks 26
Correlation of Protein Interaction Networks (PINs) and associated Metabolic
InteractionNetworks(MINs) 31
CorrelationofmetabolicfluxescarriedbyenzymesandtheirProteinInteraction
Networkproperties 34
Physical interactions in high.throughput catabolic pathways and synthesis
pathwaysofcomplexmetabolites 35
CentralproteinsinthefPIN 37
3.3Discussion 40
3.4Conclusions 43
3.5MaterialsandMethods 44
ProteinInteractionNetworks(PINs) 44
MetabolicInteractionNetworks(MINs) 45
Topologicalpropertiesofnetworks 45
CorrelationofMetabolicandProteinInteractionNetworks 47
Treatmentofmulti.enzymecomplexes 48
Thecentralityofnodes 48
CorrelationofPINsandmetabolicfluxrates 49
i
Contents
Metabolicpathways 49
Chapter4
TopologyofPhosphorylation.Networks 51
4.1Background 51
4.2Results 52
4.3Discussion 55
4.4Methods 57
Chapter5
ClassificationusingSupportVectorMachines 59
5.1SupportVectorMachines 59
5.2DimensionreductionviaPrincipalComponentAnalysis(PCA) 62
5.3Assessingandcomparingclassificationalgorithms 63
Chapter6
Detectionandcharacterizationof3D.signaturephosphorylationsitemotifs
andtheircontributiontowardsimprovedphosphorylationsiteprediction 65
6.1Background 66
6.2Results 69
Characterizationofthespatialenvironmentofphosphorylationsites 71
Computationalpredictionofphosphorylationeventsusing3D.information 77
6.3Discussion 79
6.4Methods 82
Creationofphosphorylationsitedatasets(phosSets) 82
Creationofnon.phosphorylationsitedatasets(non.phosSets) 82
Constructionofthephylogenetictreeofserine.kinases 83
Generalstructuralpropertiesofphosphorylatedandunphosphorylatedsites 83
Calculation of spatial amino acid propensity profiles, Radial Cumulative
Propensity(RCP)plots 83
Predictionapproach,evaluationofpredictionperformance 84
Feature.vectors(FV)fortheimplementedSupportVectorMachines 85
ComparisontoNetPhos,Disphos1.3andKinasePhos2.0 86
ComparisontoNetPhos,Disphos1.3andKinasePhos2.0judgedbyaccuracy 87
Chapter7
PhosPhAt:AdatabaseofphosphorylationsitesinArabidopsisthalianaand
aplant.specificphosphorylationsitepredictor 89
7.1Background 90
7.2Results 91
Databaseoverview 91
TheArabidopsispSerpredictor 94
ii
Contents
Genome.scalepredictionofphosphorylationsites 95
7.3Discussion 96
7.4Methods 97
Chapter8
Assessmentoffalsepositiveratesofphospho.proteomicdata 101
8.1Background 101
8.2Results 101
Concordanceofexperimentalreports 101
Correlation of the confidence values and the number of publication reports as
wellasthecomputedAUC 103
8.3Discussion 104
8.4Methods 105
Concordanceofexperimentalreports 105
Correlationofconfidencevaluesandnumberofpublicationreportsaswellasthe
computedAUC 105
GeneralDiscussion 107
Conclusions 115
GlossaryandAbbreviations 117
Bibliography 121
AppendixA. A.1
GOTermsusedforidentifyingProteinDegradation/Ubiquitinassociatedproteins A. 1
GOAnnotationsusedforidentifyingKinase/Phosphataseproteins A.3
GOAnnotationsusedforidentifyingDNA.relatedproteins A.4
GOAnnotationsusedforidentifyingother,non.metabolicproteins A.7
Currencymetabolites,cofactorsremovedfromthemetabolicnetwork A. 9
AppendixB. A.11
Distributionofcorrelationsaccordingtotheincludeddistances A.11
AppendixC. A.13
Detectedphysicalinteractionofenzymesinvolvedinselectedpathways A. 13
iii
Publications
Parts of this thesis have been published in peer.reviewed journals. Chapter 3
containsanintegrativeanalysisofmetabolicandproteininteractionnetworks,published
inBMCSystemsBiology , Chapter6,whichdealswitchthespatialcharacterizationand
prediction of phosphorylation sites, has been submitted to BMC Bioinformatics, while
Chapter7containsthepublicationofthe PhosPhatdatabaseinNucleicAcidsResearch .
Partsoftheoriginalversionofthelatterpublication,inwhichIhavenotbeeninvolved,
havebeenremovedfromthechapter,butretainingconsistency.Furthermore,resultsof
Chapter 8 dealing with concordance of experimental results has been accepted for
publicationinareviewofplantphosphoproteomicsin Proteomicsnextyear.
1
Abstract
Thestudyofbiologicalinteractionnetworksisacentralthemeinsystemsbiology.
Here,weinvestigatecommonaswellasdifferentiatingprinciplesofmolecularinteraction
networks associated with different levels of molecular organization. They include
metabolic pathway maps, protein.protein interaction networks as well as kinase
interactionnetworks.
First,wepresentanintegratedanalysisofmetabolicpathwaymapsandprotein.
protein interaction networks (PIN). It has long been established that successive
enzymaticstepsareoftencatalyzedbyphysicallyinteractingproteinsformingpermanent
ortransientmulti.enzymecomplexes.Inspectinghigh.throughputPINdata,ithasbeen
shownrecentlythat,indeed,enzymesinvolvedinsuccessivereactionsaregenerallymore
likelytointeractthanotherproteinpairs.Inthisstudy,weexpandedthislineofresearch
to include comparisons of the respective underlying network topologies as well as to
investigate whether the spatial organization of enzyme interactions correlates with
metabolicefficiency.Analyzingyeastdata,wedetectedlong.rangecorrelationsbetween
shortest paths between proteins in both network types suggesting a mutual
correspondence of both network architectures. We discovered that the organizing
principlesofphysicalinteractionsbetweenmetabolicenzymesdifferfromthegeneralPIN
of all proteins. While physical interactions between proteins are generally dissortative,
enzymeinteractionswereobservedtobeassortative.Thus,enzymesfrequentlyinteract
withotherenzymesofsimilarratherthandifferentdegree.Enzymescarryinghighflux
loads are more likely to physically interact than enzymes with lower metabolic
throughput. In particular, enzymes associated with catabolic pathways as well as
enzymesinvolvedinthebiosynthesisofcomplexmoleculeswerefoundtoexhibithigh
degrees of physical clustering. Single proteins were identified that connect major
componentsofthecellularmetabolismandhencemightbeessentialforthestructural
integrityofseveralbiosyntheticsystems.
Besides metabolic aspects of PINs, we investigated the characteristic topological
propertiesofproteininteractionsinvolvedinsignalingandregulatoryfunctionsmediated
by kinase interactions. Characteristic topological differences between PINs associated
with metabolism, and those describing phosphorylation networks were revealed and
showntoreflectthedifferentmodesofbiologicaloperationofbothnetworktypes.From
acloserinspection ofphosphorylation networks,weconcludedthatphosphorylationof
kinases by other kinases primarily serves to transduce signals to the ultimate target
proteinwithaparticulareffectorfunctionandinadirectedfashionbyway of forming
kinase cascades rather than to offer regulatory capacities, for example via feedback
loops. Instead, regulation was found predominantly at the level of the target protein
itself,whoseactivityappearstobefrequentlymodulatedbyseveralincomingkinases.
3
The construction of phosphorylation networks is based on the identification of
specifickinase.targetrelationsincludingthedeterminationoftheactualphosphorylation
sites (P.sites). The computational prediction of P.sites a