ChemometricsandMultivariateResolutionanditsApplicationinanalysisofTCM中南大學(xué)中藥現(xiàn)代化研究中心梁逸曾Chemometrics

Chemometricsisanewchemicaldisciplinethatusesthetheoryandmethodsfrommathematics,statistics,computerscienceandotherrelateddisciplinestooptimizetheprocedureofchemicalmeasurement,andtoextractchemicalinformationasmuchaspossiblefromchemicaldata.Chemometricscouldbedefinedasadisciplineoffundamentaltheoryandmethodologyofchemicalmeasuring.化學(xué)計(jì)量學(xué)運(yùn)用數(shù)學(xué)、統(tǒng)計(jì)學(xué)、計(jì)算機(jī)科學(xué)、以及其他相關(guān)學(xué)科的理論與方法，優(yōu)化化學(xué)量測(cè)過程，并從化學(xué)量測(cè)數(shù)據(jù)中最大限度地獲取有用的化學(xué)信息，可以說是一門化學(xué)量測(cè)的基礎(chǔ)理論與方法學(xué)。BriefHistoryofChemometricsSvanteWolduseditfirstlyforapplyingscientificprojectin1970inSweden;WoldandKowalskifoundedtheInternationalChemometricSocietyin1974;AnalyticalChemistrypublishedspecialreviewon“Chemometrics”everytwoyearsince1978;Twospecialchemometricinternationaljournalsnamed“J.Chemom.”and“ChemLab”appearedin1987frombothAmericanandEurope.ContentsinchemometricsChemometricsconsistsoffundamentalandmethodologyofchemicalmeasurements.NecessaryfundamentalknowledgeofstatisticsandlinearalgebraVectorandMatrixIsMathematicsreallyusefulforchemists?Dataexploding;Extractchemicalinformationfromthedata;RevolutionofInformationtechnique;ProgressinComputerability;VectorinanalyticalchemistryAllthespectra,chromatogramsandetc.canbenumeratedintoagroupofnumbers,whichiscalledavectorinmathematics.

HyphenatedInstruments,suchasHPLC-DAD,GC-MS,GC-IR,HPLC-MSTwo-waydatacontainingbothchromatographyandspectra;Datamatrixwithmorethan10Megabytes;DatabaseoflotsofchemicalstandardsThemixturespectrumoftwodifferentchemicalcompoundsaandbaccordingtotheLambert-BeerlawGeometricsenseofvectoradditionGeometricsenseofvectorsubtractionDirectionandlengthofvectorThedirectionofavectorisdecidedbyallitselements,sincethedifferentratiosbetweenthemcandefinesdifferentdirectionsinlinearsubspace;Thelengthofavectorisalsodecidedbymagnitudesofitselements,sincewehave,||a||=(a12+….+an2)1/2differentratiosbetweenthemcandefinesdifferentdirectionsinlinearsubspaceSubtractionoftwovectorsdefinesthedistancebetweenthetwopointsinndimensionalspaceNumericalmultiplicationofvectorsThespectraofdifferentconcentrationsInnerproductandouterproductbetweenthevectors

InnerordotproductbetweentwovectorsproducinganumberGeometricsenseofinnerproductbetweentwovectorsInnerproductandprojectionbetweenvectorsOuterproductbetweentwovectorsproducingabilinearmatrix,whichisofspecialimportanceinmultivariateresolutionWhatisthechemicalmeaningofrankofamatrix?Linearlyindependent?Rankofamatrix=thenumberofthechemicalcomponentsinthemixture?Mixturenumberandcompoundnumber?Apartofarealtwo-waydatameasuredfor

CortexCinnamomi(肉桂)

Lambert-BeerLawTheproblemhereforchemiststosolveisthat,withthemeasurementmatrixathand,oneneedstofindout:Thenumberofabsorbingchemicalcomponents:AThespectrumofeachchemicalcomponent:si(i=1,2,…,A)(firststepof

qualifications)Theconcentrationprofileofeachchemicalcomponent:ci(i=1,2,…,A)(firststepfor

quantification)Isitpossible?Yes,ifwehavethetwo-waydata!!MixturespectraandcompoundspectraThisisatwocompoundsystem;Tworedvectorsarethecompoundspectra;Sevenbluevectorsaremixturespectra;

LinearlyindependentThetwospectraarefromtwodifferentchemicalcompounds,sotheyareindependentwitheachotherSevenspectraareallfromthespectraofthetwochemicalcompounds,sotheyaredependentupontheconcentrationsofthetwocompoundsinthemixtureMixturenumberandcompoundnumber

Rankofamatrix=thenumberofthechemicalcomponentsinthemixture?

Unchangedarechemicalcompounds;Rankofamatrixisalsocertain,whichisnotchanged;Thus,ifwecollectallthespectratoformamatrix,therankofwhichshouldbe2.Howcanwefindtherankofamatrixwithmeasurementnoise?Whatarenoisesandhowdotheyinfluencethedataanalysis?Thus,weneedstatisticsandalsoalgebra;Usefultechnique:Principalcomponentanalysis!!Thisisreallyadifficultproblem!Now,let’sgothroughthisproblemstepbystep.SinglevaluedecompositionQuestion2Wehaveadatamatrixathand,weneedtoknowtherearehowmanychemicalcomponentsinit.Whattechniquedoyouwanttouse?Principalcomponentanalysis?Canweusesingularvaluedecompositiontosolvetheproblem?MultivariatecalibrationandmultivariateresolutionThisisthemajortopicinanalyticalchemometrics,wewilldiscussitinsomedetail.SinglevaluedecompositionandnumberofchemicalcomponentinthesystemLocalfactoranalysisandZerocomponentregionsEvolvingfactoranalysis(EFA)Fix-sizedmovingwindowevolvingfactoranalysis(FSMWEFA);Eigenstructuretrackinganalysis(ETA)EvolvinglatentprojectionsEvolvingFactorAnalysisEvolvingfactoranalysisFix-sizedMovingWindowEvolvingFactorAnalysis(FSMWEFA)Fix-sizedMovingWindowEvolvingFactorAnalysis(FSMWEFA)SomeproblemswithlocalfactoranalysisLocalrankZerocomponentregions,WindowsizeEigenstructuretrackinganalysisRankmapandlocaldatastructureEvolvinglatentprojectionsPrincipalComponentanalysisandLatentprojectionsEvolvinglatentprojectionsSelectiveinformationandEvolvinglatentprojections

ZerocomponentregionandnoiselevelAsimpleexampleResolutionintopurechromatogramsandspectraBecauseA=CStIfwehavepurespectraS,thenwecanusethefollowingequation:

C=AS(StS)-1Resolutionintopurechromatogramsandspectra美國(guó)標(biāo)準(zhǔn)局16組分PAH混合物標(biāo)樣(Sulpeco)

已知峰9為苯并[a]蒽和屈，峰14為苯并芘和二苯并蒽的二組分重疊峰Peaks5and6intheplot峰5、峰6的演進(jìn)特征投影圖

峰5的前5個(gè)特征值依次為16382，2436，1294，22，11分辨所得的芴、苊、菲、蒽的色譜與光譜J.CGiddingsin“Statisticaltheoryofcomponentoverlapinmulticomponentchromatograms”claimedthatItisshownthat,relativetothemaximumpeakcontentorpeakcapacityforcloselyspacedpeaks,arandomchromatogramwillnevercontainmorethanabout37%ofitspotentialpeaksand,worstofallfromananalyticalpointofview,18%ofitspotentialsingle-componentpeaks.Thenumberofobservedpeaksisnot,then,thesameasthenumberofdistinctchemicalcomponents.Thelossofanalyticalinformationresultingfromthisoverlapisbyitselfserious,buttheseverityoftheproblemisgreatlymagnifiedifwedonothaveagoodestimateofthemagnitudeoftheloss.

Anal.Chem.1983,55,418-424

ApplicationtoenvironmentalchemistryPAHsrepresentaclassoforganicpollutantsthatarewidelyfoundintheenvironment.Theyhavebeenshowntoexertaverystrongcarcinogenicactiononanimals.ThisisanexampleforqualitativeandquantitativeanalysisofPAHsinair-borneparticulatesinHongKongcitybythechemometricresolutionmethod.

1.8.2

HELP分辨

直觀推導(dǎo)式演進(jìn)特征投影法充分利用了二維數(shù)據(jù)的色譜信息和波譜化學(xué)信息，以逐步推導(dǎo)的方式分辨各組分的純色譜和波譜。具體步驟簡(jiǎn)述如下：

1）將響應(yīng)矩陣X沿色譜方向分解為一系列相對(duì)簡(jiǎn)單的子體系Xi，檢測(cè)各個(gè)子體系Xi存在的背景與噪聲，扣除背景及噪聲

2）利用演進(jìn)特征投影圖和特征結(jié)構(gòu)跟蹤分析與秩圖確定其組分?jǐn)?shù)以及各組分的選擇性區(qū)域和零濃度區(qū)域

3）利用各組分的選擇性區(qū)域和零濃度區(qū)域分辨出各組分的純色譜和波譜

4）將解析所得的各組分的純色譜和波譜乘積線性加和與實(shí)際數(shù)據(jù)比較，驗(yàn)證解析結(jié)果的可靠性

Figure2.6’TheamplifiedtotalioncurrentchromatogramoffractionN6obtainedfromGC-MS.

1.9定性分析

用質(zhì)譜庫(kù)相似檢索定性，結(jié)合色譜保留指數(shù)確正質(zhì)譜的定性結(jié)果

色譜峰B分辨所得六種物質(zhì)定性，它們依次是：6-methoxy-2-methyl-beta-carboline(C13H12N2O)、3,3’-dimethyl-[1,1’-biphenyl]-4,4’-diamine(C14H16N2)、2,5-dimethyl-4-[3-amino-4-methylphenyl]pyridine(C14H16N2)5-acetyl-4-amino-2-methylthio-thiophene(C8H8N2OS2)、4,9-dimethyl-naphtho-[2,3-b]-thiophene(C14H12S)、1,2,5,6-tetramethylacenaphthylene(C16H16)

依此原理，分辨七個(gè)餾分的所有色譜峰，對(duì)各個(gè)組分進(jìn)行定性分析。在1200多組分中鑒別出130多種含氮化合物

化學(xué)計(jì)量學(xué)與中藥現(xiàn)代化研究中草藥的現(xiàn)代化與國(guó)際化是勢(shì)不可擋的世界潮流美國(guó)自30到60年代，幾乎停止植物藥的研究（Interestinmedicalplantsprogressingsignificantlydiminishedinthe1930sandbotancalsfellintoalmostcompletedisuseuntil1960s,RoyUpton,HerbalistExecutiveDirector,AmericanHerbalPharmacopoeia），到現(xiàn)在（2000）

GlobalHerbalMarket:US$19.6billion!!（196個(gè)億美金）；德國(guó)和法國(guó)聯(lián)合開發(fā)銀杏葉提取物EG6761，找到最佳銀杏提取物的指紋圖譜，導(dǎo)致植物藥的在歐美重新流行HealthProducts(WesternCountries)GinggoBiloba

(銀杏)St.Jone’sWort(貫頁連翹)HealthProducts(WesternCountries)Greentea(綠茶)Echinacea(紫錐花)TraditionalChineseMedicine(TCM)HealthProductsDongQuai(當(dāng)歸)Ginseng(人參)TraditionalChineseMedicine(TCM)HealthProductsCordyceps(冬蟲夏草)Galic(大蒜)NaturalProductMarketGlobalHerbalMarket2000:US$19.6billion!!

NaturalProductMarket

缺乏適當(dāng)或可接受的評(píng)價(jià)傳統(tǒng)草藥的研究方法Despiteitsexistenceandcontinueduseovermanycenturies,anditspopularityandextensiveuseduringthelastdecade,traditionalmedicinehasnotbeenofficiallyrecognizedinmostcountries.…

Thequantityandqualityofthesafetyandefficacydataontraditionalmedicinearefarfromsufficienttomeetthecriterianeededtosupportitsuseworldwide.

Thereasonsforthelackofresearchdataareduenotonlytohealthcarepolicies,

butalsotoalackofadequateoracceptedresearchmethodologyforevaluatingtraditionalmedicine.傳統(tǒng)植物化學(xué)方法研究中藥的窘境每個(gè)化學(xué)物質(zhì)都分離了（耗時(shí)），但藥效似乎都不強(qiáng)，找到活性單體藥無法與同類西藥相比；中藥的君臣佐使，協(xié)同作用難以體現(xiàn)；日本人、德國(guó)人（銀杏）的發(fā)現(xiàn)，指紋圖譜的提出；中藥現(xiàn)代化需要新思路，新方法。中草藥現(xiàn)代化與國(guó)際化研究中的幾個(gè)難點(diǎn)問題中草藥的化學(xué)基礎(chǔ);(多組分復(fù)雜混合體系)中藥復(fù)方制劑;(“膏丹丸散,神仙難辯”);中藥國(guó)際化的主要難點(diǎn)：中藥基地軌范化(GAP和GEP)，中藥生產(chǎn)標(biāo)準(zhǔn)化(GMP)，基于化學(xué)指紋圖譜的中藥質(zhì)量控制;中藥作用機(jī)制研究（君臣佐使,協(xié)同作用）中藥藥效及藥代動(dòng)力學(xué)方法的建立;中藥譜（化學(xué)指紋圖譜）效（藥理藥效）學(xué)的創(chuàng)立;中醫(yī)藥理論科學(xué)化的艱巨任務(wù)廣西玉林廣東肇慶越南云南中藥指紋圖譜在中藥現(xiàn)代化研究中的核心地位EssentialConstituentsofRamulusCinnamomi(桂枝）

中藥現(xiàn)代化的機(jī)遇與挑戰(zhàn)基于中藥指紋圖譜的質(zhì)量控制、基于中藥指紋圖譜的藥效研究、基于中藥指紋圖譜的藥代動(dòng)力學(xué)研究、餾分組合（雞尾酒方法），新型更有效復(fù)方制劑的發(fā)現(xiàn)與開發(fā)、全依賴于中藥指紋圖譜的快速定性定量解析！中藥色譜指紋圖譜的測(cè)定方法以解決中藥長(zhǎng)期處于“丸散膏丹，神仙難辨”的被動(dòng)局面InstrumentsCommonlyusedforChemicalAnalysisofChineseMedicineThinLayerChromatography(TLC)(薄層色譜)

qualitativeanalysis&semi-quantitativeanalysisHighPerformanceLiquidChromatography(HPLC)(高效液相色譜)

Bothqualitativeanalysis&quantitativeanalysisGasChromatography(GC)(氣相色譜) Bothqualitativeanalysis&quantitativeanalysisDNAAnalysis(DNA分析)CapillaryElectrophoresis(CE)(毛細(xì)管電泳)中藥指紋圖譜剖析的新工具聯(lián)用色譜儀器HyphenatedInstrument(聯(lián)用儀器)Mostmodernandadvancedanalyticalinstrument.

e.g.HPLC-DAD,GC-MS,CE-DAD,etc.Combinationoftwoormoreinstrumentse.g. HPLC DAD(diode-arraydetector)

GC MS(Mass-spectrometer)SeparationInstrumentSignal

InstrumentGetmoredata/informationHPLC-DAD3DchromatogramHPLCchromatogramofnucluesideofCordycepsSinensis(冬蟲草)atonewavelengthHyphenatedInstrument(聯(lián)用儀器)HyphenatedInstrument(聯(lián)用儀器)GCinstrumentGCchromatogramof

pepticpowder(平胃散)Massspectrumtakenatretentiontime10.2minutesHyphenatedInstrument(聯(lián)用儀器)Advantages:Moredataisobtainedataretentiontime.Morespectralinformationacquiredtogivethreedimensional(3D)data.MuchmoreinformationavailabletoanalyzecomplicatedsystemlikeTCM.HyphenatedInstrument(聯(lián)用儀器)Disadvantages:Hugeamountof3Ddataisobtained.Needdataprocessingmethodsforinformationextraction,patternrecognition,etc.Computerpowerwaspoorbefore.Usualpractice:UseafewdataobtainedfromHPLC-DAD/GC-MSto findmarkercomponentsoractiveingredients.

Now,everythingbecomespossiblewithhightech.!

國(guó)家藥品監(jiān)督管理局付局長(zhǎng)任德權(quán)先生在現(xiàn)代化中藥產(chǎn)業(yè)關(guān)鍵技術(shù)系列研討會(huì)的發(fā)言中指出：

實(shí)際上，從分析化學(xué)學(xué)科自身發(fā)展看，純品的分析現(xiàn)已無特殊困難，而對(duì)復(fù)雜有機(jī)混合物體系給出快速定性定量分析，...。從現(xiàn)代化學(xué)計(jì)量學(xué)的觀點(diǎn)看，儀器分析數(shù)據(jù)的傳統(tǒng)處理方法“如光譜、波譜只取其峰值，對(duì)色譜只計(jì)算其面積等，這樣做的后果只能造成大量有用信息的浪費(fèi)。”面對(duì)形形色色的多組分體系，在充分利用化學(xué)量測(cè)儀器所產(chǎn)生的化學(xué)信號(hào)特點(diǎn)的基礎(chǔ)上，化學(xué)計(jì)量學(xué)運(yùn)用統(tǒng)計(jì)學(xué)和應(yīng)用數(shù)學(xué)及計(jì)算機(jī)，最大限度地從中抽取不同樣本中的定性、定量化學(xué)信息。在這方面我國(guó)學(xué)者梁逸曾先生按對(duì)被分析的多組分體系的定性組成知悉程度，提出了白色、灰色和黑色多組分體系三種分析體系?，F(xiàn)代分析化學(xué)學(xué)科，特別是化學(xué)計(jì)量學(xué)的發(fā)展，為中成藥指紋圖譜用于定性定量分析，提供了現(xiàn)代科學(xué)技術(shù)基礎(chǔ)。……歷史的挑戰(zhàn)與機(jī)遇并存,我們則無旁貸,充分利用時(shí)代的條件,去解決時(shí)代的問題.開創(chuàng)打破“神仙難辯”,逐步實(shí)現(xiàn)中藥定性定量分析的新時(shí)代.中藥指紋圖譜的定性定量剖析聯(lián)用色譜與化學(xué)計(jì)量學(xué)TraditionalChinesemedicines我們對(duì)二十幾種單味藥和復(fù)方制劑進(jìn)行了HPLC-DAD和GC-MS分析;對(duì)某些單味藥的道地性、季節(jié)性影響也進(jìn)行了系統(tǒng)研究，以得到它們的mono-graphs；對(duì)我省正清制藥集團(tuán)的魚腥草和風(fēng)痛寧注射液指紋圖譜的建立進(jìn)行了系統(tǒng)的方法學(xué)研究。Chemometrics:CordycepssinensisChromatogramsofthefungalpartofCordycepssinensisatwavelength(a)234nm,(b)260nmandthelarvapartofCordycepssinensisatwavelength(c)234nm,(d)260nm.AsimpleexampleChemometrics:Cordycepssinensis(a)Theselectivechromatogramsofthefunguspartfrom11.0to14.0minuteswithanintervalof5nmand(b)showsthespectraofthefunguspartintherangeof195to312nmwithanintervalof0.1minute.(c)and(d)showthecorrespondingeigenvaluesplotandthelatentprojectgraphof.(e)and(f)showtheresolvedchromatogramsandspectraforcomponentsf1andf2inthefungalpart.Chemometrics:CordycepssinensisTheHELPmethodwasusedforresolvingthesechromatograms.TheoverallresolvedchromatogramofthefungalpartTheoverallresolvedchromatogramofthelarvalpartFanGong,Yizengliang,F.t.Chau,Anal.Lett.,33（2000）2105-2128GanFengandYizengLiang,AnalyticalScience,16(2000)603-607ChengjianXu,YizengLiangandJianhuiJiang,AnalyticalLetters,33(2000)2105-2128HailinShen,YizengLiang,O.K.KvalheimandR.Manne,ChemometricsandIntelligentLaboratorySystems,51(2000)49-59(inEnglish)HailinShen,L.Stordrange,R.Manne,O.M.KvalheimandYizengLiang,ChemometricsandIntelligentLaboratorySystems,51(2000)37-47(inEnglish)HailinShen,XiaoningLi,YizengLiang,ChineseScienceBulltin（科學(xué)通報(bào)）,45(2000)587-592HailinShen,Youqunsong,HuiCui,YizengLiang,ActaChimicaSinica（化學(xué)學(xué)報(bào)）,58(2000)438-442QingsongXu,YizengLiangandKaitaiFang,ChemometricsandIntelligentLaboratorySystems,52(2000)155-166.Pepticpowder PepticpowderisanancientconcentratedpreparationinChina.

Itcanclearawaypathogenicdampness,moisturizespleen,promotethecirculationofqiandregulatethestomach.Theformulationiscomposedoffoursingleherbs.TheyareRhizomaAtractylodis(atractylodesrhizome),PericarpiumCitriReticulatae(tangerinepeel),CortexMagnoliaeOfficinalis(magnoliabark)andRadixGlycyrrhizae(licoriceroot).Thevolatileconstituentsofpepticpowderarepharmacologicalactive.GongFan,Y.Z.Liang,HuiCui,F.T.Chau,BennyT.P.Chan,Determinationofvolatilecomponentsinpepticpowderbygaschromatography-massspectrometryandchemometricresolution,J.ChromatographyA,905（2001）193-205(inEnglish)GongFan,Y.Z.Liang,Qing-SongXu,F.T.Chau,Gaschromatography-massspectrometryandchemometricresolutionappliedtodeterminationofessentialoilsinCortexCinamomi,J.ChromatographyA,909（2001）237-247(inEnglish)龔范,梁逸曾,宋又群，彭源貴,崔卉,A.K.M.Leung,Foo-timChau，高等學(xué)?；瘜W(xué)學(xué)報(bào),2001(inpress).GanFeng,Jia-hongYangandY.Z.Liang,Liberarysearchofmassspectrawithanewmatchingalgorithmbasedonsubstructuresimilarity,AnalyticalScience,17(2001)635-638(inEnglish)Y.Z.LiangandO.M.Kvalheim,Resolutionoftwo-waydata:theoreticalbackgroundandpracticalproblemsolvingPart1:theoreticalbackgroundandmethodology,FreneniusAnal.Chem.,370(2001)694-704.(inEnglish)MinghaoZhangandYizengLiang,Analyst,

(2001)(inpress).GongFan,Y.Z.Liang,Qing-songXu，F(xiàn).T.Chau,King-manNg,Anal.Chim.Acta,2001(inpress).(inEnglish)Y.Z.Liang,K.T.Fang,Q.S.Xu,ChemometricsandIntelligentLaboratorySystems,2001(inpress).

ChengjianXu,Y.Z.Liang,You-QunSongandJi-shanLi,FreneniusAnal.Chem.,(2001)(inpress)

fingerprintsforsomeherbalmedicinestToppart:(蒼術(shù)揮發(fā)油);Lowerpart:(厚樸揮發(fā)油)FindingFingerprint:HELPGC-MSofpepticpowder(平胃散)pepticpowder(平胃散)Therearefourherbsinpepticpowder:sayRhizomaAtractylodis(atractylodesrhizome),PericarpiumCitriReticulatae(tangerinepeel),CortexMagnoliaeOfficinalis(magnoliabark),andRadixGlycyrrhizae(licoriceroot).Theresultsobtainedsofarshowedthatthevolatileoilinpepticpowderarealmostthemixtureoftheabovefourherbsproportionaltotheirrelativeamountstakeninthepreparation.

FindingFingerprint:

ChemicalCompositionApproachATCMpreparation:Si-wudecoction(四物湯)100gRadixAngelicaeSinensis,當(dāng)歸80gRhizomaChuanxiong,川芎120gRadixPaeoniaeAlba,白芍

120gRadixRehmanniaePraeparata,熟地VolatileoilextractionAccordingtothestandardextractionmethodinChinesePharmacopoeiaFindingFingerprint:

ChemicalCompositionApproachRhizomaChuanxiong川芎RadixAngelicaeSinensis當(dāng)歸FindingFingerprint:

ChemicalCompositionApproachPreliminaryresults

ComponentsfoundComponentsidentifiedRelativecontentRadixAngelicaeSinensis當(dāng)歸805092.09%RhizomaChuanxiong

川芎1278284.09%Si-wudecoction

四物湯