時間序列形式的基因芯片數(shù)據(jù)的聚類分析.pdf

1、浙江大學(xué)碩士學(xué)位論文時間序列形式的基因芯片數(shù)據(jù)的聚類分析姓名：徐偉申請學(xué)位級別：碩士專業(yè)：概率論與數(shù)理統(tǒng)計指導(dǎo)教師：張幗奮201105AbstractInrecentyears，therearemoreandmorestudiesaboutmicroarrayItplaysallimportantroleinthediagnosisandtreatmentofdisease，drugscreening，selectionandbreed

2、ingofgoodcrops，andSOonAccordingtodifferentresearchpurposes，genechipdataanalysismethodsincludedifferentialgeneexpressionanalysis，clusteringanalysisanddiscriminantanalysisClusteringanalysisofgenechips，istoanalyzetherelatio

3、nshipbetweenthegeneorsample，andgeneswithsomesimilaritywouldbeclassifiedasaclassThispaperdiscussesclusteringgeneexpressiondatagivenintermsoftimeseriesTraditionalclusteringmethodsincludehierarchicalclustering，kmeanscluster

4、ingandselforganizingmapsThesemethodsbasedonconventionalsimilaritymeasuresarenotsuitableforclusteringtimeseriesdataAfewmethodshavebeenproposedrecentlyforclusteringmicroarraytimeseries，whichtakethetemporaldimensionofthedat

5、aintoaccountTheinherentprinciplebehindthesemethodsistoeitherdefmeasimilaritymeasureappropriatefortemporalexpressiondataorpre—processthedatainsuchawaythatthetemporalrelationshipsbetweenandwithinthetime。seriesareconsidered

6、duringthesubsequentclusteringphaseThispaperintroducesamethodbasedonvariationsThemethodistoanalyzetrendsofexpressionprofilesbasedontheirvariationsbetweenadjacenttimepointsItusespiecewiselinearfunctionsornaturalcubicspline

7、functionstorepresentthevariationvectorandtheintegratedsquarederrorbetweenthetwofunctionsrepresentsthedistanceofthetwoprofilesAtlast，hierarchicalclusteringisusedtoclusterprofilesExperimentsonsimulationdataandwell—knowndat