BMC Bioinformatics Review Current approaches to ge

BioMedCentralPage1of22(pagenumbernotforcitationpurposes)BMCBioinformaticsOpenAccessReviewCurrentapproachestogeneregulatorynetworkmodellingThomasSchlitt1andAlvisBrazma*2Address:1DepartmentofMedicalandMolecularGenetics,King'sCollegeLondonSchoolofMedicine,8thfloorGuy'sTower,LondonSE19RT,UKand2EuropeanBioinformaticsInstitute,EMBL-EBI,WellcomeTrustGenomeCampus,CambridgeCB101SD,UKEmail:[e-mail];AlvisBrazma*[e-mail]*CorrespondingauthorAbstractManydifferentapproacheshavebeendevelopedtomodelandsimulategeneregulatorynetworks.Weproposedthefollowingcategoriesforgeneregulatorynetworkmodels:networkpartslists,networktopologymodels,networkcontrollogicmodels,anddynamicmodels.Herewewilldescribesomeexamplesforeachofthesecategories.Wewillstudythetopologyofgeneregulatorynetworksinyeastinmoredetail,comparingadirectnetworkderivedfromtranscriptionfactorbindingdataandanindirectnetworkderivedfromgenome-wideexpressiondatainmutants.Regardingthenetworkdynamicswebrieflydescribediscreteandcontinuousapproachestonetworkmodelling,thendescribeahybridmodelcalledFiniteStateLinearModelanddemonstratethatsomesimplenetworkdynamicscanbesimulatedinthismodel.IntroductionMostcellularprocessesinvolvemanydifferentmolecules.Themetabolismofacellconsistsofmanyinterlinkedreactions.Productsofonereactionwillbeeductsofthenext,thusformingthemetabolicnetwork.Similarly,sig-nallingmoleculesareinterlinkedandcross-talkbetweenthedifferentsignallingcascadesformsthesignallingnet-work.Andthesameistrueforregulatoryrelationshipsbetweengenesandtheirproducts.Allthesenetworksarecloselyrelated,e.g.theregulatorynetworkisinfluencedbyextracellularsignals.Buttherearecharacteristicfea-turesinthesignallingnetwork,whichdonotexistintheregulatorynetwork;thereforedealingwiththesenetworksseparatelymakessense.Ourmaininterestisintranscrip-tionregulationnetworksandwewillrefertothemasgenenetworks,butmanyprinciplesarevalidforawiderangeofnetworks.High-throughputtechnologiesallowstudyingaspectsofgeneregulatorynetworksonagenome-widescaleandwewilldiscussrecentadvancesaswellaslimitationsandfuturechallengesforgenenetworkmodelling.Thissurveyislargelybasedonandisanexten-sionoftwopreviouspublications[1,2].Genenetworksareconcernedwiththecontroloftran-scription,i.e.howgenesareupanddownregulatedinresponsetosignals.Inthe1960'sgeneticandbiochemicalexperimentsdemonstratedthepresenceofregulatorysequencesintheproximityofgenesandtheexistenceofproteinsthatareabletobindtothoseelementsandtocontroltheactivityofgenesbyeitheractivationorrepres-sionoftranscription.Theseregulatoryproteinsarethem-selvesencodedbygenes(Figure1).Thisallowstheformationofcomplexregulatorynetworks,includingpos-itiveandnegativefeedbackloops.Theseprinciplesofgeneregulationapplytoprokaryotes(e.g.bacteria)aswellastoeukaryotes(e.g.higherorganisms).ThecontrolofgeneactivityismuchmorecomplexthanFigure1suggests.Itinvolvesmanykindsofproteinsthusallowingadditionallevelsofcontrolparticularlyineukaryotes.Transcriptionfactors,theproteinsthatrecognizetheregulatoryele-Published:27September2007BMCBioinformatics2007,8(Suppl6):S9doi:10.1186/1471-2105-8-S6-S9supplementtitlepOttoWarburgInternationalSummerSchoolandWorkshoponNetworksandRegulation/p/titleeditorPeterFArndtandMartinVingron/editornoteReviews/note/supplementThisarticleisavailablefrom:©2007SchlittandBrazma;licenseeBioMedCentralLtd.ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.BMCBioinformatics2007,8(Suppl6):S9(pagenumbernotforcitationpurposes)mentsintheDNA(thebindingsites)needtointeractwithotherproteinsinordertoactivategeneexpression.Inadditiontocontrolofgeneexpressionthereareregulatorycontrolstodeterminethematuration,transportanddeg-radationofthemRNA,aswellasitstranslation.Justtoillustratethecomplexityofgeneregulation:GeneOntol-ogy(GO),acontrolledvocabularyusedtodescribepro-teinfunctionscontainscurrentlyover7500differenttermsdescribingbiologicalprocess'transcription',includ-ingover6500termsunderprocess'regulationoftran-scription'[3].Genenetworksareoftendescribedverballyincombina-tionwithfigurestoillustratesometimes-complicatedinterrelationsbetweennetworkelements.Duetothecom-plexityofthesenetworks,suchmodelsarenotalwayseasytocomprehendandtheyoftenleaveaconsiderableamounttoambiguitytothereader'simagination.Sincethe1960'smethodsfrommathematicsandphysicshavebeenusedtodescribeandsimulatesmallgenenetworksmorestringently.Nowadays,molecularbiologicalmeth-odsandhigh-throughputtechnologiesmakeitpossibletostudyalargenumberofgenesandproteinsinparallelena-blingthestudyoflargergenenetworks.Thisallowstack-linggenenetworksmoreefficientlyandhasledtoanewdisciplinecalledSystemsBiology,whichseekstocombinemethodsfrombiologywithmethodsfrommathematics,physicsandenginee