2

3A.PEARSONIfyouweretomakeascatterplotofthevaluesofxagainsty(pairingx1withy1,x2withy2,etc),thenrreportshowwellyoucanfitalinetothevalues. y D=1-4SpearmanrankcorrelationKendall’sτThesemethodsaremorerobustagainst56789B.EuclideanUnlikethecorrelation-baseddistancemeasures,theEuclideandistancetakesthemagnitudeofchangesinthegeneexpressionlevelsintoaccount,Itshouldthereforeonlybeusedforexpressiondatathataresuitablynormalized.A.HIERARCHICALHierarchicalclusteringmethodsorganizesgenesor/andsamplesinatreestructure,basedontheirThebasicideaistoassembleasetofitems(genesorarrays)intoatree,whereitemsarejoinedbyveryshortbranchesiftheyareverysimilartoeachother,andbyincreasinglylongerbranchesastheirsimilaritydecreases.A.HIERARCHICALB.K-MEANSChoosekinitialcenterpointsClusterdatausingEuclideandistance(orotherdistanceCalculatenewcenterpointsforeachclusterusingonlypointswithintheclusterRe-ClusteralldatausingthenewcenterRepeatsteps3&4fordefined聚類分析-1.3R

一 一

一個(gè)變量如果是由大量微小的、獨(dú)立的隨量的疊加結(jié)果，那么這個(gè)變量就服從正態(tài)分布。因此許多隨量都可以用正態(tài)分布表示或 >[1]>[1]>1-[1][1]55538

[,1] 3Fisher'sExactTestforCountDatadata:countsp-value=alternativehypothesis:trueoddsratioisnotequalto195percentconfidenceinterval:2.124677oddsratioGOaccession: ption:cellproliferationTermtype: Is“cellproliferation”overrep