文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > 自然科学 > 生物学 > A unified mixed-model method for association mapping
侵权投诉

A unified mixed-model method for association mapping

LETTERS

a

Percentage100b

Percentage806040

20©2006 Nature Publishing Group doc.guandang.net kinshipRelative kinshipFigure2DistributionofpairwiserelativekinshipestimatesintheCEPHsampleandmaizesample.Thepeaksaround0indicatenorelationship.(a)FortheCEPHsample,thesmallpeaksaround0.25representrelationshipsamongoffspringwithinafamily,parent-offspringorgrandparent-parent,andthesmallpeakaround0.125representrelationshipsofgrandparent-offspring.(b)Forthemaizesample,thecontinuouslydescendingestimatesofrelativekinshipagreewiththecomplexfamilialrelationshipsandpopulationstructureamongthesemaizeinbredlines.Forsimplicity,weshowonlypercentagesoftherelativekinshipestimatesofthemaizesamplefrom0to0.45.Association(SA)andGenomicControl(GC)arecommonmethodsusedinhumanandplantstudiestocontrolforthefalsepositives(typeIerrors)causedbythispopulationstructure9–13.WithGC,randommarkersareusedtoestimateandadjustthein ationofteststatisticsgeneratedbypopulationstructure,assumingsuchastructurehasasimilareffectonallloci.SAanalysisusesrandommarkerstoestimatepopulationstructureandthenincorporatesthisintofurtherstatisticalanalysis.ForsamplesinareaIII,however,accountingonlyforpopulationstructuremayleadtoeitherinadequatecontrolforfalsepositivesoralossinpowerowingtofamilialrelatedness(Fig.1).Itremainstobeseenwhichmethodswillprovemostusefulwhenevaluatingsampleswithveryhighlevelsofpopulationstructurealongwithdiverselevelsoffamilialrelatedness(areaV;Fig.1).Inthispaper,wepresentanewmethodforassociationmappingthatisapplicabletosamplesfromareasIIandIII(Fig.1).Forthefamily-basedsample,weappliedourmethodtomicroarraydatafromthebaselineexpressionlevelsofgenesinimmortalizedBcellsfrom14familiesofCentred’EtudeduPolymorphismeHumain(CEPH)Utahpedigree14.Inthisstudy,sixgeneexpressionphenotypeswereregard-edasphenotypictraitsinmappingexpressionquantitativetraitloci

(eQTL).Forthesamplecontainingcomplexfamilialrelationshipsand

A unified mixed-model method for association mapping

A unified mixed-model method for association mapping

populationstructure,weexaminedthreequantitativetraitsmeasured

on277diversemaizeinbredlines,representingthediversitypresentin

publicbreedingprogramsaroundtheworld15.Asmaizeisahighly

outbredspecies,thepopulationdifferentiation(Fst)amongthemajor

subgroupsinoursamplerangedfrom0.047(SSR)to0.073(SNP),

similartothatofarecenthumanstudy11(Fst¼0.013forChinese-

JapaneseandFst¼0.145forbetweencontinents).Whenweincluded

aminorbottleneckedsubgroup,theoverallFstroseto0.106(SSR)

and0.118(SNP).

Ourassociationmappingapproachintegratesgenomictoolsto

uncoverpopulationstructureandfamilialrelationshipswiththetraditionalmixed-modelframeworkthathaslongbeenusedbyanimalgeneticists16–18.Oneobviousobstacleinapplyingamixed-modelmethodbeyondafewdomesticatedanimalspeciesisthatpedigreerecordsareoftenunknownorinaccurate.Genomictoolsnowallowustodetectbothpopulationstructure(Q)andrelativekinship(K)withinasample.Marker-basedrelativekinshipestimateshaveprovenusefulforquantitativeinheritancestudiesindifferentpopulations19,20.ThisKestimateapproximatesidentitybydescentbyadjustingtheprobabilityofidentitybystatebetweentwoindividualswiththeaverageprobabilityofidentitybystatebetweenrandomindividuals.FortheCEPHsample,wereplacedthepedigree-basedcoancestrymatrix(G)inatraditionalmixedmodelwiththeKmatrixtode nethedegreeofgeneticcovarianceamongindividuals.Nopopulationstructurewasdetected,andQwasnotincludedinthemixed-modelanalyses.Forthemaizesample,we tbothQandKintothemixedmodeltoaccountformultiplelevelsofrelatedness.Werandomlytestedtheexpressionlevelofsixgeneswithgoodheritabilityestimates:HSD17B12,TUBB2A,CTSH,RPS26,UBE2L3andSSR1.TheKmatrixagreedwiththefamilystructureofthedata.Although94%ofthepair-wisekinshipestimateswerecloseto0,thesmallpeaksaround0.25and0.125representedtherelationshipswithinfamilies(Fig.2a).ThemixedmodelwitheithertheKorGmatrix tsthedataequallywell.TheKmodelshowedasigni cantimprovementinmodel toverthesimplemodelinwhichfamilystructureisignored(Table1).Overall,theKmodel,GCandQTDTshowedgoodcontroloftypeIerrorrate(Fig.3a–candSupplementaryFig.1online).TheKmodelgaveslightlyliberalresultsandGCgaveslightlyconservativeresultsforTUBB2A,whereastheresultsfromQTDTwereliberalforTUBB2AbutslightlyconservativeforHSD17B12.Thestatisticalpowersimulationwasconductedbyaddingageneticeffecttoeachmarkerandthentestingwhetheritcouldbedetectedbydifferentmodels.TheadjustedTable1Goodnessof tofthreedifferentmodelsinexplainingphenotypicvariationofhumangeneexpressions

HDS17B12

–2log

likelihood

Simplemodel

Coancestry

Kmodel

aModelTUBB2A–2loglikelihood210.3a174.9173.9CTSH–2loglikelihood208.1a137.7140.7RPS26–2loglikelihood184.9a56.558.7UBE2L3–2loglikelihood72.3a43.645.0SSR1–2loglikelihood14.0a–3.2–3.2BIC126.685.883.6BIC220.8190.7189.7BIC218.6153.5156.5BIC195.472.374.5BIC82.859.460.8BIC24.612.612.6116.1a70.067.8comparisonbasedonw2testindicateswhethertheKmodelsigni cantlyimprovesthemodel tatPo0.001;BIC,BayesianInformationCriterion(smallerisbetter);allCEPHUtahfamilymemberswereusedintheanalysesofthreemodels;thesimplemodelwasincludedonlyforthepurposeofillustratingtheeffectofignoringfamilyrelationships,asitisnotastandardpractice.NATUREGENETICS

上一页第2页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email