代谢组学介绍

Introductionofmetabonomics/metabolomics2009-06-26Theflowofthe“omics”sciences:genomics,proteomics,andmetabolomicsSpratlinJ.L.,etal.ClinCancerRes,2009January15,15(2):431-440What’sinaname?Metabolome“…referstothecompletesetofsmall-moleculemetabolites(suchasmetabolicintermediates,hormonesandothersignallingmolecules,andsecondarymetabolites)tobefoundwithinabiologicalsample,suchasasingleorganism…”Oliveretal.,1998代谢组“是指基因组的所有下游产物也即最终产物的组合，这些产物是一些参与生物新陈代谢、维持生物体正常功能和生长发育的小分子化合物，主要是相对分子量小于1000Da的内源性小分子”许国旺著.代谢组学-方法与应用,科学出版社，2008年第一版:第一章,P1-10Metabonomics“…measurementofthedynamicmultiparametricmetabolicresponseoflivingsystemstopathophysiologicalstimuliorgeneticmodification…”Nicholsonetal.,1999Metabolomics“...thecompletesetofmetabolites/low-molecular-weightintermediates,whicharecontextdependent,varyingaccordingtothephysiology,developmentalorpathologicalstateofthecell,tissue,organororganism…”Oliver2002代谢组学“是通过考察生物体系（细胞、组织或生物体）受到刺激或扰动后（如将某个特定的基因变异或者环境变化后），其代谢产物的变化或其随时间的变化，来研究生物体系的一门科学”许国旺2008What’sinaname?Analyticalplat-forms：(1)Nuclearmagneticresonance(NMR);(2)GasChromatography–MassSpectrometry(GC-MS);(3)LiquidChromatography-MassSpectrometry(LC-MS);etc.GC-MSLC-MSMetadataobtainTaoX.M.,etal.AnalBioanalChem.,2008,391:2881-2889TotalionchromatogramDataobtain(1)Filteringandpeakdetection滤噪、峰检测(2)Deconvolution重叠峰解析（3）Peakalignment峰对齐、匹配（4）Normalization归一化Dataanalysisandinterpretation(5)非监督的模式识别方法：利用获取的样本信息，对样本进行归类，并采用相应的可视化技术直观的表达出来，不需要有关样品分类的任何背景信息。该方法将得到的分类信息和这些样本的原始信息（如疾病的种）进行比较，建立代谢产物与这些原始信息的联系，筛选与原始信息相关的标志物，进而考察其中的代谢途径。常用的非监督学习方法如主成分分析(principalcomponentsanalysis),系统聚类分析…主成分分析的基本思想：对变量X进行线性变换，形成新的综合变量PC；根据实际需要选择2-3个PC进行分析，以达到降维和简化问题的作用（多元二元/三元）PC1=a11X1+a21X2+…+ap1XpPC2=a12X1+a22X2+…+ap2Xp许国旺等著.代谢组学-方法与应用,科学出版社，2008年第一版:第12章,146-156PCAscoresplotofonsetALLandAMLpatientsDataanalysis(6)有监督的模式识别方法：利用一组已知分类的样本作为训练集，让计算机对其进行学习，获取分类的基本模型，进而可以利用这种模型对另一组分类未知的样本进行类别识别。常用的有监督学习方法如偏最小二乘判别分析(Partialleastsquares-discriminantanalysis，PLS-DA)，正交偏最小二乘判别分析，费舍尔线性判别分析…许国旺等著.代谢组学-方法与应用,科学出版社，2008年第一版:12,146-156偏最小二乘法分析思想对变量进行分类：设定p个因变量Y1,…,Yp和m个自变量X1,…，Xm，对两类变量进行建模。提取自变量的第一成分T1和因变量的第一成分U1，使T1和U1相关程度达最大,然后建立U1和T1的回归方程；如果回归方程未达到满意的精度，则用同样的方法提取T2和U2。T1=w11X1+…+w1mXmT2=w21X1+…+w2mXm判别分析思想应变量为定性变量，且分组类型在两组以上；自变量为可测量的度量变量。计算（线性）判别式；将自变量代入判别式，计算每个观察样本的判别Z得分，然后根据得分值对其进行归类。t(1)t(2)Thescorest,onevectorforeachmodeldimension,arenewvariablescomputedaslinearcombinationsoftheX's.TheyprovideasummaryofXthatbothapproximateXandpredictY.PLS-DAscoresplotofonsetALLandAMLpatientsOtherstatisticapproaches,suchasttestandANOVA,arealternativesatthisstep.VIP(variableimportanceintheprojection)valuesTheinfluenceofeveryterminthematrixXonalltheY's.VIPisnormalizedsothatSum(VIP)2=K(numberoftermsinthematrixX).TermswithVIP1haveanaboveaverageinfluenceonY.DeviationofeachvariablesfromALL(standarddeviationsfromaverage)Potentialbiomarkeridentification:standardStudent’sttestorANOVABlindpredictiontestofPLSDAmodelY-PredictedThreemajorstepsofmetabolomicsanalysisSpratlinJ.L.,etal.ClinCancerRes,2009January15,15(2):431-440Clinicalapplicationsofmetabolomicsinoncology1.SearchearlydiagnosticbiomarkersBreastcancer:tChoglycerophosphocholineglucose2.Responseassessmenttochemicaldrugs/therapytreatmentsBothasapredictivemeasureofefficacyandapharmacodynamicmarkerTizianiS,LodiA,KhanimFL,ViantMR,BunceCM,etal.PLoSONE,2009,4(1):e4251BathenTF,etal.BreastCancerResTreat,2007;104:181–189.Someknowledgeaboutprostatecancer1.Prostatecancer—themostfrequentlydiagnosedcancerinmen2.currentdiagnosticmethods:usingacombinationofdigitalrectalexaminationandmeasuringthelevelsoftheenzymePSAinthebloodserum3.limitationofcurrentdiagnosis:thefeaturesofthiskindofcancerarenotoriouslyvariableamongpatients.MetabolomicprofilingofprostatecancerScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914a,Venndiagramofthetotalmetabolitesdetectedacross42prostate-relatedtissuesand110matchedplasmaandurinesamples.b,Venndiagramof626metabolitesintissuesmeasuredacross16benignadjacentprostatetissues,12clinicallylocalizedprostatecancers(PCA)and14metastaticprostatecancers(Mets)ScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914MetabolomicprofilingofprostatecancerScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914HierarchicalclusteranalysisofprostatetissuesamplesScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914bluecircles-benignadjacentprostateyellowsquares-localizedprostatecancerredtriangles-metastaticprostatecancerPrincipalcomponentsanalysisofprostatetissuesamplesScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914blue-benign;yellow-localizedtwo-tailedWilcoxonranksumtestScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914yellow-localized;red-metastaticScreekumarA.,etal.Nature,2009Feruary12,457(7231):910-914Aroleforsarcosineinprostatecancercellinvasionandandrogensignaling