生物信息學(xué)中模體聚類算法的研究.pdf

1、分類編。：密級(jí)：公丌中位代碼：10065號(hào)：09209023又繂砷泛大嗲研究生學(xué)位論文論文題目：生物信息學(xué)中模體聚類算法的研究StudyonmotifclusteringalgithmsinBioinfmatics學(xué)生姓名：陳超中請(qǐng)學(xué)位級(jí)別：理學(xué)碩士.請(qǐng)專業(yè)名稱：計(jì)算機(jī)應(yīng)用技術(shù)研究方向：牛物信息學(xué)指導(dǎo)教師姓名：張少強(qiáng)專業(yè)職稱：副教授提交論文日期：2011年3月31日滅津師范人學(xué)碩I:學(xué)位論文ABSTRACTCurrently

2、inbioinformaticsclusteringalgithmshavealreadybeenwidelyusedinmoLiffindinghoweversomeofthoseapproachesaredoingwellwhenbindingsitesaresignificantlydifferentfromtheirbackgroundsequences.Butinfactsomemotifsmayhavearelatively

3、similarnucleotidedistributiontothatoftheirbackgroundsequences.Sounderthissituationitiswthwhileenoughtotestcomparethosemethodsviausingrealsyntheticdatasets.Thispaperfirstlyintroducesanumberoftoolsaimingtoclusterdatasets.M

4、eoverIanalyzedcomparedtheirperfmancebyusingsyntheticrealdatasetsaseriesoftestingstrategies.Allthetestsfocusedontheirexecutingspeedandaccuracyoftheirresults.Ididextensivepreparatywkbefecarryingoutthetests.Allthedataedwash

5、pickedpurposefully.Todothissuccessfully，Idesignedanintegrateddetailedprocedurefselecting.Igavearelativelydetailedintroductionofthosemethodsaswellasthetests.Theexperimentsaimingatcomparingthemusedthevariousdatasetsincludi

6、ngthosehavearelativelysimilarnucleotidedistributiontothatoftheirbackgroundsequences.Bydoingsoitmadethealltestsmeclosetotherealisticcondition.Imptantlythosemotifwouldbemissedbysometools.Afterfinishingthetestswhichemployed

7、bothsyntheticandrealdatasetsfmotiffindingIdiscoveredthateachofalgorithmshasitsownadvantagesinsomeareas.AmongthemCliClusteringshowedanoutstingperfmanceinparticularfbalancingthepredictionsensitivityspecificityingeneral.Iti

8、smelikelytoidentifythebindingsitesthantheothertoolswhenthedistributionofnucleotidesofbindingsitesissimilartothatoftheirbackgroundsequences.Meoveritiswellcomplementarywithothertools.Keywds：clusteringmethodsbioinfmaticsmot