sjtul matlab8 v02生物信息學(xué)第八課

LifeScienceNecessities:FlexibilityandDataGathering–BreadthandEasilyloadcommonfileExcel,CSVandotherImage(jpeg,tiff,gif,bmp,png,AccesstomanyspecializedSequencedata(fasta,embl,genbank,etc.)Microarray(Affymetrix,GenePix,GEO,BLASTReports,MassSpec,PhylogeneticTrees,CompleteintegrationtoSQLandODBCDirectAccesstoExternalVideoCameras,MedicalEquipment,Example:SeamlessDatabaseVisualQueryAccessdatawithoutknowingScrollthroughtablesandCustomizeyourBuilt-invisualizationPlottingandCreatingHMTLHandlingdateReuseSQLstatementsinyourownProblemswithinsufficientlyautomatedcomputationalLackofInadequatemetricsforquantification,Slow,Humanerror,transcriptionLimitedscientificPerformaspectrumofanalysesincludingnonlinearmixed-effects（非線性混合效應(yīng)）,sequence（測(cè)序）,microarray（微陣列）,phylogenetictree（系統(tǒng)進(jìn)化樹）,massspectrometry（質(zhì)譜分析）,andgeneontology（基因本體論）Importdatafrommultiplesources,suchasdatabases,fileformats,orShareresultswithautomaticallygeneratedHTMLreports,datavisualizations,orstand-alonetoolsParallelizedataanalysistodecreasecomputationAutomateanalysestoimplementbatchprocessingofcontiguousExploreProductsforComputationalDataAcquisitionTheadvantageofautomatedcomputationalObtainobjectiveReducecostsandDecreaseprocessingandanalysisAlleviatehumanerrorsandtranscriptionConsiderthisimagefromNationalCancerGoal:ToquantifytheamountofInitialmethod:Post-docsitsbehindmicroscopeandcountsthenumberofmetastaticspotsnottootimeconsumingforoneimage

NotaveryconvincingGoal:ToquantifytheamountoftissuemetastasisforInitialmethod:Post-docsitsbehindmicroscopeandcountsthenumberofmetastaticspotsHowautomatedcomputingObtainobjectiveReducecostsandDecreaseprocessingandanalysisAlleviatehumanerrorsandtranscriptionTectorialMembraneGoal:DetermineelasticityofTectorialMembraneAtomicForceMicroscopeInitialmethodtoAtomicForceMicroscopeAnalysisof1AFMfiletook30-40Arealisticgoalwastoanalyze10filesinoneWithautomatedcomputing,theobtainableamountofdataincreasedAnalysisof1AFMnowtook3-4Nowwecouldanalyze100soffilesinaportionofaAnalysisofFluoresceinGoalDeterminemeancirculationtime平均循環(huán)時(shí)間(MCT)andretinalbloodflow視網(wǎng)膜血流Intensity, Intensity,Fit ensity-vs-TimetolognormalparameterizedbyIo,Ip,tp,b(shapeMCTMCT=tm,vein-RBF=2art+ AnalysisofFluorescein =t-t)exp3 Manuallytrackvessels,collectingtime-intensitydata(40minutesinadarkroom!)Manuallyidentifyarteries,TransferintensityinformationtostatisticspackagetocalculatefitparametersDetermineManuallymeasurevesselCalculateLogresultsinlabPerfectapplicationforneuralnetworksAutomatedtheanalysiswithMATLABandCodecurrentlyusedinlabsLet’stakeaGoal:Determinemeancirculationtime(MCT)andretinalbloodflowPreviousTimeVeryAutomatedcomputingallowedusObtainobjectiveDecreaseprocessingandanalysisReducecostsandTypicalAccessAnalyzeShareSimBiology,Systems

SimBiology?providesanappandprogrammatictoolstomodel,simulate,andanalyzedynamicsystems,focusingonpharmacokinetic/pharmacodynamic(PK/PD)andsystemsbiologyapplications.Itprovidesablockdiagrameditorforbuildingmodels,oryoucancreatemodelsprogrammaticallyusingtheMATLAB?language.SimBiologyincludesalibraryofcommonPKmodels,whichyoucancustomizeandintegratewithmechanisticsystemsbiologymodels.Avarietyofmodelexplorationtechniquesletyouidentifyoptimaldosingschedulesandputativedrugtargetsincellularpathways.SimBiologyusesordinarydifferentialequations(ODEs)andstochasticsolverstosimulatethetimecourseprofileofdrugexposure,drugefficacy,andenzymeandmetabolitelevels.Youcaninvestigatesystemdynamicsandguideexperimentationusingparametersweepsandsensitivityanalysis.YoucanalsousesinglesubjectorpopulationdatatoestimatemodelSimBiologyUserinterfacetofacilitatebuilding,simulating,andanalyzingdynamicImport,build,andexportmechanisticorPKPDrepresentationofsystemSimulateresponsestobiologicalvariabilityordifferentdosingconditions,scanparameterranges,calculatesensitivitiesLeast-squaresestimationofgroupedorpooleddata,andmaximumlikelihoodestimationofpopulationparametersDeploySimBiologymodelsforstandaloneQuestionstoWhatisthevalueofmodelingQuestionstoWhyCreateQuantitativeBiochemicalReactionBiochemicalpathwaysstartoutsimpleandquicklygrowinTestingpathwaysviaexperimentisexpensiveinbothtimeandmoney.QuantitativemodelingnarrowstherangeofOncecreatedandvalidatedwithexperimentsthequantitativemodelcanbeusedasanin-silicosandboxtotestnewideasdramaticallyfasterthanthroughexperimentation.ChallengeswithincomputingbiochemicalIntegratingknowledgefromexperimentaldata,intuition,literature,andothermodelsisdifficultModelersandscientistshavedifficultycommunicatingknowledgeandsharingworkThemathematicsforsolvingthesemodelsisevolvingfasterthanthetoolsManydifferenttoolsareneededtocompleteentireworkflowModelcreatedbyEnterinchemicalEstimateparametersusingexperimentaldataIsolaterelevantparametersusingsensitivityanalysis>>IntroductiontoProvidesoneenvironmentforbothgraphicalandprogrammaticIntroductiontoProvidesonetoolformodeling,simulating,andanalyzingpathwaysUsedbymodelersorprogrammerstogaininsightintotheirpathwayandtocommunicatetheirpathwaywithKeyBuildingaTabularViaMATLABImportSBMLRunningaAnalyzingaSensitivityLet’sLet’sbuildasimpleAsimplegeneregulationmodelwithtranslation,andnegativefeedbacktosuppressLet’sbuildasimpleTranscription:theprocessthroughwhichaDNAsequenceisenzymaticallycopiedbyanRNApolymerase聚合酶toproduceacomplementaryRNA;thetransferofgeneticinformationfromDNAintoRNA.Translation:thesecondpartofproteinbiosynthesis生物合成,inwhichanmRNAsequenceisconvertedtoachainofaminoacidstoformaprotein.

>>>>Pharmacokinetics.Thestudyofwhatthebodydoestoadrugafteradministration.是指抗生ThestudyofAbsorptionDistributionMetabolismandExcretion分泌(ADME)ofdrugsinthebodyPharmacodynamics.Thestudyofwhatthedrugdoestothebody.是指抗生素在感染部位達(dá)到相應(yīng)的濃Thestudyofthebiochemicalandphysiological生理學(xué)effectsofdrugsmechanismsofdrugactionrelationshipbetweendrugconcentrationandeffectPROBLEM:Theeffectofadrugiscalculatedfromtheamountinthebiophase,which,unfortunately,cannotbedirectlymeasured.PKknowledgeisneededtomodeltransferofdrugfrombloodtoeffectsiteChallengesinPK/PDManytoolsChallengesinPK/PDNONMEM,Basic,Fortan,C:Buildingandmaintainingmodelscanbedifficult.OrganSpecificornicheSimulationtoolsaretoocomplexand/orblackboxOrganmodelsnoteditable,methodsarenotFlexibilityisWorkflowismanual,notModelling,simulation,statistics,andvisualizationallrequiredifferenttoolsManualintegrationistimePKExampleTransdermalInputNicotinepatchisappliedtotheskinfor16Overlappingzero-orderinputDrugconcentrationmonitoredfor24Singlecompartment

Rapiddecreaseinconcentrationwheninfusionratesdrop==

Totaldose–Doseslow

dC/dt=(FfastdC/dt=(Ffast+Fslow–

NoPKExample…PKExample…1234568FastinfusionrunsfortimeSlowinfusionrunsfortimeInitialnicotineconcentration=2V=140V=140=78 =6=17

GenericPBPKmodelofFromPoulinandThiel;JPharmaceuticalSciences.91:5,MayFromPoulinandThiel;JPharmaceuticalSciences.91:5,MayPKExample–Let’sPKExample–Let’sshowhowwemightimplementthisin>>>>Read,analyze,andvisualizegenomicandproteomicBioinformaticsToolbox?providesalgorithmsandappsforNextGenerationSequencing(NGS),microarrayanalysis,massspectrometry,andgeneontology.Usingtoolboxfunctions,youcanreadgenomicandproteomicdatafromstandardfileformatssuchasSAM,FASTA,CEL,andCDF,aswellasfromonlinedatabasessuchastheNCBIGeneExpressionOmnibusandGenBank?.Youcanexploreandvisualizethisdatawithsequencebrowsers,spatialheatmaps,andclustergrams.Thetoolboxalsoprovidesstatisticaltechniquesfordetectingpeaks,imputingvaluesformissingdata,andselectingBioinformaticsToolbox--KeyNextGenerationSequencinganalysisandSequenceanalysisandvisualization,includingpairwiseandmultiplesequencealignmentandpeakdetectionMicroarraydataanalysis,includingreading,filtering,normalizing,andMassspectrometryanalysis質(zhì)譜分析includingclassification,andmarkerPhylogenetictreeGraphtheoryfunctions,includinginteractionmaps,hierarchyplots,andpathwaysDataimportfromgenomic,proteomic,andgeneexpressionfiles,includingSAM,FASTA,CEL,andCDF,andfromdatabasessuchasNCBIandGenBankThemicroarraydataforthisexampleisDeRisi,J.L.,Iyer,V.R.,andBrown,P.O.(Oct24,1997).Exploringthemetabolicandgeneticcontrolofgeneexpressiononagenomicscale.Science,278(5338),680–686.PMID:9381177.TheauthorsusedDNAmicroarraystostudytemporalgeneexpressionofalmostallgenesinSaccharomycescerevisiaeduringthemetabolicshiftfromfermentationtorespiration.Expressionlevelsweremeasuredatseventimepointsduringthediauxicshift.ThefulldatasetcanbedownloadedfromtheGeneExpressionOmnibusWebsiteat:1、LoaddataintotheMATLABenvironment.loadyeastdata.mat2、GetthesizeofthedatabyAns=Accesstheentriesusingcellarray%Thisdisplaysthe15throwofthevariableyeastvalues,whichcontainsexpressionlevelsfortheopenreadingframe(ORF)YAL054C.ans=4、UsethefunctionwebtoaccessinformationaboutthisORFintheSaccharomycesGenomeDatabase(SGD).url=5、AsimpleplotcanbeusedtoshowtheexpressionprofileforthisORF（openreadingframe）.xlabel('Time(Hours)');6、Plottheactualvalues.plot(times,2.^yeastvalues(15,:))xlabel('Time(Hours)');ylabel('RelativeExpressionLevel');TheMATLABsoftwareplotsthefigure.ThegeneassociatedwiththisORF,appearstobestronglyup-regulatedduringthediauxicshift.7、Compareothergenesbyplottingmultiplelinesonthesamefigure.holdxlabel('Time(Hours)');ylabel('RelativeExpressionLevel');title('ProfileExpressionLevels');TheMATLABsoftwareplotstheThisprocedureillustrateshowtofilterthedatabyremovinggenesthatarenotexpressedordonotchange.Thedatasetisquitelargeandalotoftheinformationcorrespondstogenesthatdonotshowanyinterestingchangesduringtheexperiment.Tomakeiteasiertofindtheinterestinggenes,reducethesizeofthedatasetbyremovinggeneswithexpressionprofilesthatdonotshowanythingofinterest.Thereare6400expressionprofiles.Youcanuseanumberoftechniquestoreducethenumberofexpressionprofilestosomesubsetthatcontainsthemostsignificantgenes. M‘emptySpots=strcmp('EMPTY',genes);yeastvalues(emptySpots,:)=[];genes(emptySpots)=[];2、Usetheisnanfunctiontoidentifythegeneswithmissingdataandthenuseindexingcommandstoremovethegenes.nanIndices=any(isnan(yeastvalues),2);yeastvalues(nanIndices,:)=[];genes(nanIndices)=[];ans3、UsethefunctiongenevarfiltertofilteroutgeneswithsmallvarianceovertimeThefunctionreturnsalogicalarrayofthesamesizeasthevariablegeneswithonescorrespondingtorowsofyeastvalueswithvariancegreaterthanthe10thpercentileandzeroscorrespondingtothosebelowthethreshold.mask=%Usethemaskasanindexintothevaluestoremove%filteredyeastvalues=yeastvalues(mask,:);genes=genes(mask);ans4、Thefunctiongenelowvalfilterremovesgenesthathaveverylowabsoluteexpressionvalues.Notethatthegenefilterfunctionscanalsoautomaticallycalculatethefiltereddataandnames.[mask,yeastvalues,genes]=ans5、Usethefunctiongeneentropyfiltertoremovegeneswhoseprofileshavelowentropy:[mask,yeastvalues,genes]=ans H=?ln(1/30)= uniformNowthatyouhaveamanageablelistofgenes,youcanlookforrelationshipsbetweentheprofilesusingsomedifferentclusteringtechniquesfromtheStatisticsandMachineLearningToolbox?1、Forhierarchicalclusteringthefunctionpdistcalculatesthepairwisedistancesbetweenprofiles,andthefunctionlinkagecreatesthehierarchicalclustertree.corrDist=pdist(yeastvalues,'corr');clusterTree=linkage(corrDist,'average');2、ThefunctionclustercalculatestheclustersbasedoneitheracutoffdistanceoramaximumnumberofclustersInthiscasethe'maxclust'optionisusedtoidentify16distinctclusters.clusters=cluster(clusterTree,'maxclust',3、Theprofilesofthegenesintheseclusterscanbeplottedtogetherusingasimpleloopandthefunctionsubplot.forc=1:16plot(times,yeastvalues((clusters==c),:)');axistightsuptitle('HierarchicalClusteringof4、TheStatisticsandMachineLearningToolboxsoftwarealsohasaK-meansclusteringfunction.Again,16clustersarefound,butbecausethealgorithmisdifferentthesearenotnecessarilythesameclustersasthosefoundbyhierarchicalclustering.forc=

TheMATLABsoftwareiterations,totalsumofdistances=iterations,totalsumofdistances=8.6267426iterations,totalsumofdistances=8.8606622iterations,totalsumofdistances=9.7767626iterations,totalsumofdistances=9.010354、TheStatisticsandMachineLearningToolboxsoftwarealsohasaK-meansclusteringfunction.Again,16clustersarefound,butbecausethealgorithmisdifferentthesearenotnecessarilythesameclustersasthosefoundbyhierarchicalclustering.forc=5、Insteadofplottingalloftheprofiles,youcanplotjusttheforc=1:16axistightaxisoff %turnofftheaxissuptitle('K-MeansClusteringofClustering6、YoucanusethefunctionclustergramtocreateaheatmapanddendrogramfromtheoutputofthehierarchicalClusteringPrincipal-componentanalysis(PCA)isausefultechniqueyoucanusetoreducethedimensionalityoflargedatasets,suchasthosefrommicroarrayanalysis.YoucanalsousePCAtofindsignalsinnoisydata.1、UsethepcafunctionintheStatisticsandMachineLearningToolboxsoftwaretocalculatetheprincipalcomponentsofadataset.[pc,zscores,pcvars]=pca(yeastvalues)

TheMATLABsoftwarepcColumns1through2、Youcanusethefunctioncumsumtoseethecumulativesumofthevariances.cumsum(pcvars./sum(pcvars)*Thisshowsthatalmost90%ofthevarianceisaccountedforbythefirsttwoprincipal

TheMATLABsoftwareans3、Ascatterplotofthescoresofthefirsttwoprincipalcomponentsshowsthattherearetwodistinctregions.Thisisnotunexpected,becausethefilteringprocessremovedmanyofthegeneswithlowvarianceorlowinformation.Thesegeneswouldhaveappearedinthemiddleofthescatterplot.xlabel('FirstPrincipalComponent');ylabel('SecondPrincipalComponent');title('PrincipalComponentScatterPlot');4、ThegnamefunctionfromtheStatisticsandMachineLearningToolboxsoftwarecanbeusedtoidentifygenesonascatterplot.Youcanselectasmanypointsasyoulikeonthescatterplot.5、AnalternativewaytocreateascatterplotiswiththegscatterfunctionfromtheStatisticsandMachineLearningToolboxsoftware.gscattercreatesagroupedscatterplotwherepointsfromeachgrouphaveadifferentcolorormarker.Youcanuseclusterdata,oranyotherclusteringfunction,togroupthepcclusters=clusterdata(zscores(:,1:2),6);xlabel('FirstPrincipalComponent');ylabel('SecondPrincipalComponent');title('PrincipalComponentScatterPlotwithColoredgname(genes)%Pressenterwhenyoufinishselectinggenes.SupportedDataSupportedDataBLAST

GeneExpressionOtherDataDesignofPrimersforAutomatedDNACalculatepropertiesofFilterprimersbasedonGCcontentorCheckfordimerizationandhairpinRetrieveprimerFindrestrictionenzymethatcutinsideIsolateprimerslackingaGC Pos 50cacatagcccttgccataag11375054.37AppliedBiosystemsDevelopsaCrucialDNASequencingAlgorithminMATLAB?TheTodeveloparobustyetflexiblecalibrationalgorithmtobeincludedinahigh-throughputDNAanalysisinstrumentTheUseMATLABtotestideasandcodeaprototype,andthenusetheMATLAB