简介：Inthepost-genomicera,variouscomputationalmethodsthatpredictprotein-proteininteractionsatthegenomelevelareavailable;however,eachmethodhasitsownadvantagesanddisadvantages,resultinginfalsepredictions.Herewedevel-opedauniqueintegratedapproachtoidentifyinteractingpartner(s)ofSemaphorin5A(SEMA5A),beginningwithsevenproteinssharingsimilarligandinteractingresiduesasputativebindingpartners.ThemethodsincludeDwyerandRoot-Bernstein/Dillontheoriesofproteinevolution,hydropathiccomplementarityofproteinstructure,patternofproteinfunctionsamongmolecules,informationondomain-domaininteractions,co-expressionofgenesandproteinevolution.AmongthesetofsevenproteinsselectedasputativeSEMA5Ainteractingpartners,wefoundthefunctionsofPlexinB3andNeuropilin-2tobeassociatedwithSEMA5A.WemodeledthesemaphorindomainstructureofPlexinB3andfoundthatitsharessimilaritywithSEMA5A.Moreover,avirtualexpressiondatabasesearchandRT-PCRanalysisshowedco-expressionofSEMA5AandPlexinB3andtheseproteinswerefoundtohaveco-evolved.Inaddition,weconfirmedtheinterac-tionofSEMA5AwithPlexinB3inco-immunoprecipitationstudies.Overall,thesestudiesdemonstratethatanintegratedmethodofpredictioncanbeusedatthegenomelevelfordiscoveringmanyunknownproteinbindingpartnerswithknownligandbindingdomains.

简介：贝叶斯分类方法,是以概率假设为基础的.假设需要分类的数据遵循某种假设分别,通过这些概率分布以及对观察数据本身进行推理,得出最好的分类结果.本文重点介绍贝叶斯分类模型的相关理论基础以及常见的几种贝叶斯分类模型.

简介：利用粗糙集的思想,基于K均值聚类算法对社会网络进行研究,聚类划分社会网络,处理社会网络聚类边界模糊问题.在社会网络中精心社区挖掘与聚类分析相似,将社会网络节点作为聚类分析对象,社会网络社区为聚类分析的类.实验结果表明,本文所提方法提高了社区划分的准确率,使聚类结果能够全面反映社会网络.