/
(Beta,unfinished)(M.Pagel@Reading.ac.uk)Meade(A.Meade@Reading.ac.uk)
. (Beta,unfinished)(M.Pagel@Reading.ac.uk)Meade(A.Meade@Reading.ac.uk)
.

(Beta,unfinished)(M.Pagel@Reading.ac.uk)Meade(A.Meade@Reading.ac.uk) . - PDF document

jane-oiler
jane-oiler . @jane-oiler
Follow
429 views
Uploaded On 2015-07-24

(Beta,unfinished)(M.Pagel@Reading.ac.uk)Meade(A.Meade@Reading.ac.uk) . - PPT Presentation

MajorIntroductionMethodsFormatRunningRunningcommandtimeMarkovtraittraitsGeneralisedSquaresforcontinuouslytraitsTestingsamplingMCMCmechanismandmixinganalysisMixing ID: 91731

MajorIntroductionMethodsFormatRunningRunningcommandtimeMarkovtraittraitsGeneralisedSquaresforcontinuouslytraitsTesting:samplingMCMCmechanismandmixinganalysisMixing.....................................

Share:

Link:

Embed:

Download Presentation from below link

Download Pdf The PPT/PDF document "(Beta,unfinished)(M.Pagel@Reading.ac.uk)..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.


Presentation Transcript

(Beta,unfinished)(M.Pagel@Reading.ac.uk)Meade(A.Meade@Reading.ac.uk) MajorIntroductionMethodsFormatRunningRunningcommandtimeMarkovtraittraitsGeneralisedSquaresforcontinuouslytraitsTesting:samplingMCMCmechanismandmixinganalysisMixing.........................................................................................................................AcceptanceRateJumpMultistateMultistateMCMCexampleexamplestatereconstructiondiscretevaluesexamplesindependentdependentJumpreduction(Model(Model (ModelMCMCsignificancekappa,EstimatingunknownExample:EstimatingunknownExample:EstimatingunknownIndependentListproblemFrequentlyReferences deviationautomaticallytunedpolytomysnewMultipleIndependentmodelinternalcontinuousmodelsbugimprovementsTreesarehavebranchdiscretemultistateestimateswillcomputerperformingamongphylogenyavailable.appliedtheadoptnumberdiscretecontinuouslytestedmodelsancestralcorrelationsThemethodintouncertaintyaboutevolutionApproachMarkovMonteCarlo(MCMC)posteriordistributionsmaximum(ML)methodspointphylogenies.TheconventionalMCMCtheMarkovdistributionmodelsevolutionparametersmodelsbelow).withexplored,appliedtreesmodelsestimatedtestedtakingphylogeneticuncertaintyintoaccount.   BayesPhylogeniespackagegeneratealignmentdata methodsMultiStatetraitsadoptnumberdiscretephylogenetictrees.usefulreconstructingancestralappliedtraitsadoptmorediscrete2004.Systematic53,usedevolutionbetweendiscretebinaryMostthebutcould“high”,features.mightincludetestsgeneticculturalthecorrelationsamongtraits,theother(see1999.models(ref)detectvariationstree,accountinglineagemanualdesignedprogramsimplementtheseDetailedaboutmethodspapersend(someavailablewebsite).SyntaxcommandsBayesTraitsNexustreesincludepolytomybutbranchlengths.includedthetreebutlinkednumberthesectionnumberexampleareincludedwithprogram.Formatfromtext(ASCII),taxonthetree.mustspelledexactlytheandthem.speciesname,firstcolumndata,thisadditionaltake“A”,“C”datamustexactlytakeintegerspoints.MultiStateremainingdatadatacontinuousremovedtree.Continuousincludedwith ExampleMultiStatedata BBC  ‐  CCB      AB datasite.datatreatedtraitcouldtheforuncertain.ThesignifiescanbutExample(binary)data  ‐  10 01     11 ExampleContinuous 2 34     1 commandpromptterminal(OSLinux),doubleclickingTheprogram,placeStartthecommandpromptterminalchangedirectorydataandtype.OSXTreeFilenamethefilenamethedata multiplecomplexcommandscommandfile,typingthemtime.commandtextfile,commandsexampleincludedThe 1 1 MultiState,Artiodactyinputthecommand.discretecontinuousmodelschangefromanygivenanyotherstaterateparametersrates1994discussion).Theestimatingtransitionratestheeachbelowtheadoptandamongandtheseestimatestree,thesefromthisindicatestheysignificantcomponentbuttheExamplemodelevolutionadoptsstatesState 2  ‐‐ q q states,thematrixtwelvebinarytraitjust forevolutionbinary(logcontinuousMarkovwhichtraitsindependentlyEachdescribedoneabove,but twoand“1”.createscoefficientspertraitsthatthechangeonedependsbackgrounddependentcanadoptfourcombinationthe(0,0;0,1;1,0;1,1).representedprogramdescribechangesotherconstant.mainestimateddiagonalelements‘dual’occur,beingthattheseratestraitschangingexactlythesameoverperiodshowever.(1994)Barker(2005)forfurtherthisState 0,11, 1,1  ‐‐ ‐‐‐‐ ‐‐‐‐  ‐‐ therateparametersuponthemeasurementphylogeny.branchfactor‘c’theratesdecreasedthefactor‘c’.parametersMarkovdiscussedmodelBayesTraitsimplements(Tuffley147:63–91,1998).variantcontinuouswithinandbetweenbranches.elegantthatattention,findlimitedwithcomparativethemodelrequiremodelContinuousanalysescontinuouslyvaryingdata(GLS)approachBrownianPagel,1997,1999).nonindependencespeciesaccountedamongspecies.thephylogeneticthechange(theBrownian‘rate’change,andancestralstaterootcancovariancecorrelation.approachmeansthatcanacrossspeciesandinterpretedtheobtainedContinuous.approachContinuouspossibletransformphylogenytheadequacyunderlying phylogeneticrequired,andaboutexample,evolutionpunctuationalgradual,adaptiveconstant.   HypothesisTesting:Factorsdoeshypothesesyoubutinformationneededtests.conjunctionwiththeexamplesthetwooftenused.(LR)likelihoodsderivednestedexpressedthatother).calculated2[logfittingdistributeddegreesequaldifferencethenumberbetweentheHowever,circumstances1994,1997Pagel,2005)thefewerincludetheInformationnotdescribetestsmanyphylogeneticinferenceexample,Phylogenies,2004).InformationthatMCMCoftenBayeslogictest,exceptthan   integralparameterspracticethislikelihoodcanapproximatedharmonicallowingMarkovrunlargenumbercheckthatharmonicBayesTraitsthetheharmonicprogramignoredduringtheperiodwhentheconvergence.Therunningharmonicmeansfromfinaliterationthevaluestheindependentanddependentthencompared.just   2(log[harmoniclog[harmonicmean(simple factor Interpretation Strong testingcontroversialanalysis,otherAIK,considered. WhenMCMCdistributionsthechosen.theparametersdependentuponbranchthingsequal,brancheswillparametersversa.thetheprior(e.g.,uniform,exponential)theuninformativepriorsareandwhenthethedataButcomparativethereonepointsmanyhundredsthousandssequence   softunderbellyanalyses.guidingprinciplechoicemustgoodthatoftenrunyourtreestheaveragetheparameters.optionconstrainparameterspriornarrowerandbiologicalgroundsperhapsbutindicationmidpoint.rulewhenconstrainedinformed(nonvaluestruncatedeitherupperlowerconstrainedrange,mustchanged.programexponential,betadistributedexponentialthenwhereastakeuniexponential.exponentialusefulgeneralthanthoughtvalue,intermediatemeanused.(uniform,beta)parametersthedistribution.exponentialmeantheparameterq01uniformtheq10uniform100thesamecandoestakeBecausedifficultdistributions,allowshyperhyperdrawntheexponentialusingelegantsometheuncertaintyarbitrarinessMCMCpriorshyperMeade,2004ancestralcharacterSystematicBiology,53,673684. Whenhyperapproachspecifyforuniformdistributionthatdistribution.forexponentialexponentialprioruniformtheHyperPriorAllmeanandfromhyperbothinterval10.MCMCrunwhileconvergence.impossiblehardmanyburnthatminimumseldomThelengthburnincommand.burnMorecomplexlargertreesperiods.iterationsMarkovthereprintingeachlineoutput.Insteadthinnedthatoutputindependent.samplecommand.printthetheautocorrelationamongcheckedevenExcel).comparativedatasets,everyadequaterun1010000default,changedcommand,whichthenumberiterationsruninfinitewhichMixing,timeschangesuccessfulanalysis.changesparameterlargelikelihooddramatically,convergencetheThistheacceptanceratebecomingbetweenotheraccepted.Thesmallchangesproposechangemuch,leadinghighidealratebetweenconvergence.valuebetweendatatrees,parametersbodymanymagnitudeobtainedtimegenomeThisverymechanism.parameterbiggerleadingloweracceptanceleadhighercontinuousparameterstreeacceptancerate,cannottuned.attemptsautomaticallytunetheproposalmechanismrateparameteracceptanceapproximatelybut RateDevcommand.Thecommandtakesnumber,commandtheMonitoringAcceptanceschedulewhichmixing,containsschedule,percentageoperatorsfollowedheader.headernumbertimesthepercentagetimetunetheacceptancethatparameter,iterationrunningacceptancerateTheschedulechainmixingcorrectly.constantbattlecomparativemethodsbetweencomplexityanddataneededmanyespeciallydiscretethemanyparameters,cannotestimatedthemodelparameterised.Indicationsinclude,estimatedparameters,parameterstradingeachsuboptimallikelihoods,convergenceModelcomplexityparametersrestrictcommandensuringratiohigh.restrictionmultistatediscretemodelsnumberparameters,manysupported.BayesTraitsnumberincreaseavailableestimateremainingones.Thecommand(res)usedrestrictcommandparameternames,suppliedparametersalpha2followingcommandindependentrestrictedconstants,theunrestricted(UnRes)usedrestrictionsModeltestingparameterjustified,numberreduced. JumpMCMCcomplexmodelnumberandimpossiblereverseMCMCwasintegratemodelparameterdescription(ref).RevJumpcommandusedjumpMCMC,commandtakethecommandexponentialTheexponential ‐  RJHP exp 0 100 Multistateprogram“Artiodactyl.trees”treefiletheThefollowingMultiStatemaximumlikelihoodoptionsdisplayinginformation.Thiswhatexpect.runThefollowedheader Output number,500this for DG GD D G outputprinted.Oncebeentheterminate. Multistateexampleprogram“Artiodactyl.trees”treeand“Artiodactyl.data”multistateMCMC(2).defaultexponentialprinted Output Iteration MeanRunningharmonic number DG GD thestate thestate Iteration No 110007.933077.933071873.7025612.54460.3510230.648977 120008.988468.593984953.1209734.9149590.4756280.524372 130008.374168.525943.7993834.4897980.4173930.582607 1400010.28069.3123117.0761327.544980.4999720.500028 1500010.71229.789126.9455885.8652190.484360.51564 …………………… 10060008.944819.903884008.976618.4075630.4735170.526483 10070008.532449.903131470.336441.4777020.0121160.987884 10080008.035629.902283382.4540933.3349730.2708690.729131 10090008.411399.901511072.2174074.1835070.3659740.634026 10100009.728129.901357.505419.5387850.4841140.515886 OutputdesignedexcelJMP.Rundependthescheduled“Artiodactyl.txt.Schedule.txt”created,checkchainmixingbetweenqGDestimatedbothqDGqGDsignificantlyfromanalysistakeqDG.used shouldbutparametersqGDtheiteration.Thecalculatingmeans.harmonicanalysisqDG9.90135,harmonicqDG8.965492.qGD,becausethantheqGD.D. mean(complexex mean(simple model)]  Log  BF = 2(‐9.901358.965492)   1.871716lesssimplerharmonicbetweendependingrandomillustrativepurposes.harmonicmeanscalculatedandtestedreliabilityusingmultipleindependentTheymodelAddMRCAAddNodecommandsancestralstatesmultistatediscretemodels.ThethetwocommandsThecommandsidentifyandlisttaxanumbernode.graphicsgeneratecommandclickingappropriate“Artiodactyl.trees”multistatecommandsdefinedFKWhaleThenodecalledidenticalbutnumbertaxadefinenode.followingcommandscolumnsP(D)”andP(G)”,thereconstructing    presentGoat,commandsreconstructionpresentbecausepresenttrees.commandreconstructstheMostRecentpresentMRCA.Anynumbercanreconstructedanalysiseffecteachnodesexternalavailableonecommandtakesnodetaxadefinenode,foundmethod.Thecommanddefinedfossilisedsignificant,state.discreterequiresnumberinsteadThetablebelowthenumberstherecorrespondingleftunchanged, ‐  Number ‐‐‐ X ‐‐‐ XX XX X XX XX X XXX0 XX0X X0XX 0XXX usedtraitssignificanceestablishedlikelihoodsmodels,independently,assumesexamplesMCMCbutused.matingwrittenindependentevolveindependently,thetraitindependentsecondtrait.independentparameters,alpah2beta2.transitions1,0 Parameter Transitions alpha1 1110 alpha2 2210 10 2 1  ‐  0 1 2 BayesTraitstheindependentSetexponentialanalysis,commands. Run willmultistateheadercontain. Output Iteration MeanRunningharmonic number transition transition P(0,0)thestate0,0 P(0,1)thestate0,1 P(1,0)thestate1,0 P(1,1)thestate1,1 dependentassumesthecorrelatedchangedependentdependentmodelq34,q43. Parameter Dependent Transitions 1=0201 2=0101 1=0210 2=1101 2=0110 1=1201 2=1110 1=1210 Q0 QQ  ‐  0QQ BayesTraitsthedependentmodel(3)analysis(2).Setexponentialanalysis,commands. Run willindependentmodelexceptthatthedependentareestimatedtheindependent.calculateFactorbetweenindependentdependentindependentsharmonicdependentharmonic2(log[harmonicmean(complexlog[harmonicmean(simple42.4745.51)   suggestsevolution.HarmonicbetweenMCMCreductionthedatacomplexitymodelunlikelyindependentanddependentmodelsestimatedThepreviousexample,demonstratedcouldsimplifiedandrestrictionssignificant.TherearepossibleindependentdependentReverseMCMCMCMC)resultsautomaticallyjumpcommandtakespriorcommandexponentialMCMCcanused RJHP exp 0 100 theprimatesandwiththedependentMCMCthecommandswillcontaincolumns. Output ParametersNumber ZeroNumber stringparameterrestrictions modeldependent(D)independent(I)    numberswhicharegroupsareexamplefordependentoneforparameter,the“'1twozero.groupparameters(q12,q42),groupformedparameters(q21,q31),parametercancheckedtheparameterdataindependentMCMCandcomparedependentMCMC.described(ref),theswitchingratebetweenThemodelthedifferenttree.commandusedthemodel,columnsincludedtheOff”ThebetweenstatesBayesTraitsthe“Mammal.trees”“MammalBody.txt”,thesampletreestherebodyillustrativepurposesSelectmodelmaximumlikelihoodstartanalysis    Output number,this for theroot corrected BayesTraitsthe“Mammal.trees”“MammalBody.txt”,Model(2), Output Iteration MeanRunningharmonic number 1 corrected continuouscorrelatedresultsdefault)Runtree“Mammal.trees”andSelectmodelMCMCtheinformationprintedheader    Output Iteration MeanRunningharmonic number 1 1 correctedthe correctedthe correlation analysisthecorrelation TestCorrel Run shouldsimilartheTraitcorrelationtestedharmonicmeanstheproducedharmonic73.29analysisharmonic135.03,wouldFactorsuggestinghighlyusedtesttraitsdistanceModelultrametrictreestipbetweentaxa.fictional“MammalModelB.txt”canusedtestsignificantdirectionaltrendpreformingModelModelRegressioncontinuoususedtestsignificanceunknownThefirstdependentMammalBrainBodyGt.txtmammalbrain,time.“MammalBrainBodyGt.txt”Select(6)andanalysis.header Output Iteration MeanRunningharmonic number Intercept coefficientfortrait coefficientfortrait motion squared squared s.e.AlphaStandardAlpha s.e.Standard s.e.Standard significancenumberwayssignificantmodel,harmonicmeanstrait.Thethethetimecoefficientcoefficientscommand.numberlambdaTheseparametersallowtempo,mode,phylogenetic threedefault.Thesethatlengthsdescribeconstantmodeltraitfollowedbranchlengths,they(e.g.,estimatingthebetweensignificantlyimprovesdatathemodel.differentiallyphylogeneticbranchpunctuationalversustraitKappabranchesshorterindicatingthatbranchescontributelongbranch).Kappa1.0longerbranchesones.theextremeKappatraitevolutionindependentlengthbranch.Kappawithpunctuationaldeltaphylogeny ‐ distancethedetecttimetips,radiations.Delta(earlierphylogeny)contributeevolutionthisradiation:changeindicatesthatcontributeThisacceleratingevolutiontimeway,thatdetectsdifferentialovertimethewhichconstant.phylogenycorrectlypredictsthepatternsamongspeciestrait.Thisparameterindicatesassumptionsunderlyingthecomparativemethodsindependenttruephylogenytrait.wereindependent,parametertakephylogeneticcanlambdatreebeingrepresentedbigphylogeny.expectedlambdamotion)beingrepresentationIntermediatethelambdadifferentphylogeny.thetraitsthenlambdaestimatedsimultaneouslyestimatingthecharacterisetraitslambdacanestimatedeach. Parameter contributionphylogenyphylogenyindependent)phylogeneticphylogenydefined branchpunctuationalbranchesbrancheschange totaltotaldefinedchange(adaptive scalingandinterpretationphylogenythreeparametersestimatedMCMC,threeeitherparameteritsestimation(theyestimateddefault,1.0)scalingparameterfollowednumberthecommandestimatescommandestimateskappa,thecommandestimatesdelta,0.5Modeltestingsignificantsignificant,lambdasignificantlyfromunknownmodelsestimateunknowninternaldata.Estimatingunknowndistributionestimatedfromdata,areunknownTheprocesspreventestimateddatamodelparameters.Estimatingwithmodelmodelthemodelbutanalysis.“SaveModels”commandspecified“LoadModels”commandmodelsThemodelspecifiedwhencreatingandestimatingunknowncheckingimplemented.unknowninternalnodesBayesTraitsthe“MammalBody.txt”SelectmodelMCMCanalysiscommands“MamBodyModels.bin” SaveModels MamBodyModels.bin Run “MamBodyModels.bin”created.estimatedata,treemodel(4)analysis(2).TheAddMRCAcommandestimateinternaltakesthenodeoutputlisttaxadefinecommandsthe“Nodedefinedtaxa,reconstructcalled“Nodedefinedtaxa,analysis. AddMRCA Node-01 Whale Hippo Llama Ruminant Pig headercontain Output Iteration MeanRunningharmonic number number 1 motion Est ‐ valuesNode Est ‐ valuesNode unknownforestimated,nodes.“MammalBrainBodyNoTapir.txt”beencreatedfortree“MammalBrainBodyNoTapir.txt”Selectandcommandmodelsanalysis Run “MammalBrainBodyPredTapir.txt”containsthebutthequestiondatausedindicateshouldestimated.and“MammalBrainBodyPredTapir.txt”datamodel(6)(2).commandbelowanalysis Run willcontain“EstTapirwithpredicted IndependentindependentmodelsML.the“Mammal.trees”SelectIndependentcontrast(7)MCMCanalysis.independentcontrastindependent.contain Output Iteration MeanRunningharmonic number 1 2 motion motiontrait modelrateschangetimewherethedifferedsignificantly,description(ref).MCMCidentifytheevolutionsignificantly.onlysingletreerequiresroughlythereincluded.anddataselectIndependentcontrast(7)MCMCruncommands VarRates Run willcontain Output Iteration MeanRunningharmonic number 1 motion VarRatesareas filescreated,“Marsupials.txt.PP.trees”themodifiedincreasedchange,havedecreasedThe“Marsupials.txt.PP.txt”detaileddescriptionchanges,formatnumberuniqueandtaxanumberinternalnodes,uniquenodebranchlengthroot),numberwhichdefinelistID.Thesectionresultsthethe Output the Priorprior Pramchangethe phylogenetic motion Scaleprior(unchanging,diagnostics changetreethere Output nodechange change thechangecreated Branchchangenodebranch designedcomputerextractinformationfromthefile.Commandcomment.Shortcut:Example:comment.ancestorShortcut:taxanumbernodereconstruct.Example:Taxa2Shortcut:taxanumbernodereconstruct.Example:AddNodeNode1AddNNode1interceptShortcut:Example: numberiterationsburnMCMCinfiniteShortcut:Example:maximumnumberShortcut:Example:CapRJRatesShortcut:Example:deviationparameterusedperturbingestimateddata.Thisnotrecommended.Shortcut:number,Example:EstimateShortcut:estimatenumberExample:EqualTreeschainThisposteriordistributiontree.Shortcut:NumberiterationsburneachExample:EqualTrees20000midpointsample.Shortcut:Example:runningtheanalysis Shortcut:Example:internalShortcut:whichdefinenode.fixingaboveinformation.Example:Hylobates_gabriellaeHylobates_leucogenysEstimategammaheterogeneity.Shortcut:numbercategoriesparameterExample:commands,Shortcut:Example:parameterShortcut:distributioneachparameterfromExample:gamma100commonShortcut:distributionparameterExample:Shortcut:Example:numberShortcut:numberchaininfinitetermination.Example:1000000 parameterShortcut:estimatenumberExample:parameterShortcut:estimatelambda,numberExample:0.8modelsmodelShortcut:Example:ModelFile.binnumbertimesmaximumlikelihoodhighertakelongerrunShortcut:NumbermaximumlikelihooddefaultExample:frequenciesShortcut:frequenciesestimates,est,unifrequencies,empirical,Example:Shortcut:distributionparameters,include,beta,gamma,Example:UniformShortcut: distributionparameters,commandExample:thenumbercategoriesdivideinto,default100.Shortcut:Example:rateparameter,effectingacceptancerate.SettingShortcut:automaticallyestimateparameter,valuenamespecificparametercontinuousExample:RestrictRestrictparameterparametersShortcut:parameterrestrictExample:Restrictalpha1Restrictalpha1alpha2Restrictbeta11.5RestrictAllRestrictparameterShortcut:Example:RestrictAllalpha1RevJumpanalysisShortcut:Example:RevJumpRevJumpGammaRevJumpHPanalysishyperShortcut:Example:RevJumpHPexpRevJumpHPgamma theShortcut:Example:Shortcut:Example:modelsShortcut:modelsExample:ModelFile.binModelFile.binShortcut:treesExample:randomShortcut:numberExample:39362SymmetricalcreatesymmetricalShortcut:Example:SymmetricalnamesnumbersShortcut:Example:correlationbetweentesting.Shortcut:Example: UnRestrictparameterrestrictionShortcut:UNResExample:q01UnRestrictAllShortcut:UnResAllExample:variablemodelShortcut:Example: FrequentlyQuestionsrunningprogramclickingprogramcommandclickingit.“Runningnexusblock.template.descriptionsnumbernamemustnotspecifiedbutnotdataCheckspellingtaxanumbersnexusbecausefile.causestreefile.mainmemoryerrorsrunningmemory,duetreescomplexintensivemodels.memoryversionprogram.runningtheprogramsmallernumbertreesThecauseerrorprogrammingpleasethecommandused.notmixingbetweenproblemcausedlikelihoodsignificantlythanthesample.likelihood,bettermuchThisproblemnumberwhenthesupported,chancecombinationcreatebetterlikelihoodpreventingmixing.problemdeterminethechaingetslikelihood.optionsthereason.secondcommandspendequalamounttree.treecommandproduceposteriorparametersintegrated References