18989.限定性主成分分析算法的建立及其初步應(yīng)用

1、分類號UDC密級編號中閹斜孥萱記砑窺生阮碩士學(xué)位論文墮塞絲圭盛坌坌塹篡洼笪建童壁蕓麴堂廑雖邱爽申請學(xué)位級別理堂殛學(xué)科專業(yè)名稱生物信息堂論文提交Et期2Q12堡壘基論文答辮鱉期2魚12生基學(xué)位授予單位主國疊堂醫(yī)班塞壘隨答辯委員會主席限定性主成分分析算法的建立及其初步應(yīng)用AbstractMicroarraytechnologyhasbeenwidelyusedtoidentifygeneswithdifferentialexpression

2、pattemsunderdifferentconditionsDifferentexpressionpattemsinexperimentalandcontrolgroupscanbemeasuredfrombothastaticanddynamicpointofviewSincetheregulationofgeneexpressionisadynamicprocess，anexperimentdesignwithtime—serie

3、sdatawillprovidemorebiologicalinformationthanastaticexperimentaldesignManymethods，includingPCA，SVD，theBayesianmodel，HMM，ANOVAandregressionbasedmodels，areusedtoanalyzetimeseriesmicroarraydatafromvariousperspectives，mostof

4、whicharedesignedtofindgenesundergoingsignificantchangesHowevertheydonottakeintoaccounttheinherentcorrelationamongdifferenttimepoints，andnoiseisalsoignoredWhatismore，fewmethodsemphasizethetimeseriespropertyunderlyingtimes

5、eriesdataInspiredbythetraditionalPCAthatuseoneaxiswithmaximizedvariancerepresentasmuchinformationaspossibleandremovethenosieandbythetimeforcedlinearmodelthatembedtimeseriespropertyamongsamplepointswhenprocessingtimeserie

6、sdata，theauthorestablishedanewmodelnamed“ConstrainedPrincipalComponentAnalysis”(CPCA)，whichattemptstofindoneaxiswhichcanmaintaintimeseriespropertyamongsamplepointsonconditionthatvarianceismaximizedtepresentthemaintrendof

7、aseriesofvectorsTheauthorusedtwosetsofpubliclyavailabletimeseriesexpressiondatatotestwhethertheCPCAalgorithmispracticabletofinddifferentiallyexpressedgenesfromtimeseriesexpressiondataunderdifferentenvironmentalconditions