Protein Homology Modelling[蛋白質同源建模](PPT-48).ppt

ProteinHomologyModellingThomasBlicherCenterforBiologicalSequenceAnalysisLearningObjectivesAfterthislessonyoushouldbeableto:Explaintheindividualstepsinvolvedincalculatingaproteinhomologymodel.Identifysuitabletemplatesformodelling.heprinciplesbehindabinitioproteinstructureprediction.Describethedifferencesbetweenhomologymodellingandabinitiostructureprediction.Describethemajorpitfallsinproteinmodelling.OutlineProteinhomologymodellingIndividualstepsCaveatsPitfallsAbinitioproteinstructurepredictionThreadingTrueabinitiomethodsWhyDoWeNeedHomologyModelling?AbInitioproteinfolding(“random”sampling):100aa,3conf./residuegivesapproximately1048differentoverallconformations!RandomsamplingisNOTfeasible,evenifconformationscanbesampledatpicosecond(10-12sec)rates.Levinthal’sparadoxDohomologymodellinginstead.HowIsItPossible?Thestructureofaproteinisuniquelydeterminedbyitsaminoacidsequence(butsequenceissometimesnotenough):prionspH,ions,cofactors,chaperonesStructureisconservedmuchlongerthansequenceinevolution.Structure>Function>>SequenceHowOftenCanWeDoIt?Therearecurrently~47000structuresinthePDB(butonly~4000ifyouincludeonlyonesthatarenotmorethan30%identicalandhavearesolutionbetterthan3.0Å).Anestimated25%ofallsequencescanbemodeledandstructuralinformationcanbeobtainedfor~50%.pletegenomesSignalingproteinsDisease-anismsMembraneproteinsProtein-ligandinteractions”Foldspacecoverage”StructuralGenomicsinNorthAmerica10year$600millionprojectinitiatedin2000,fundedlargelybyNIH.AIM:structuralinformationon10000uniqueproteins(now4-6000),sofar1000havebeendetermined.Improvecurrenttechniquestoreducetime(frommonthstodays)andcost(from$100.000to$20.000/structure).9researchcenterscurrentlyfunded(2005),targetsarefrommodelanddisease-anisms(aseparateprojectonTBproteins).HomologyModelingforStructuralGenomicsRobertoSánchezetal.NatureStructuralBiology7, 986 - 990(2000)HowWellCanWeDoIt?Sali,A.&Kuriyan,J.TrendsBiochem.Sci.22,M20–M24(1999)