ThemonoclepackagetakesamatrixofexpressionvalueswhicharetypicallyforgenesasopposedtosplicevariantsascalculatedbyCuinks3oranothergeneexpressionestimationprogramMonocleassumesthatgeneexpressionva ID: 124547
Download Pdf The PPT/PDF document "Ascellsdi erentiate,theyundergoaprocesso..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Ascellsdi erentiate,theyundergoaprocessoftranscriptionalre-conguration,withsomegenesbeingsilencedandothersnewlyactivated.Whilemanystudieshavecomparedcellsatdi erentstagesofdi erentiation,examiningintermediatestateshasprovendicult,fortworeasons.First,itisoftennotclearfromcellularmorphologyorestablishedmarkerswhatintermediatestatesexistbetween,forexample,aprecursorcelltypeanditsterminallydi erentiatedprogeny.Moreover,twocellsmighttransitthroughadi erentsequenceofintermediatestagesandultimatelyconvergeonthesameendstate.Second,evencellsinageneticallyandepigeneticallyclonalpopulationmightprogressthroughdi erentiationatdi erentratesinvitro,dependingonpositioningandlevelofcontactswithneighboringcells.Lookingataveragebehaviorinagroupofcellsisthusnotnecessarilyfaithfultotheprocessthroughwhichanindividualcelltransits.Monoclecomputationallyreconstructsthetranscriptionaltransitionsundergonebydi erentiatingcells.Itordersamixed,unsynchronizedpopulationofcellsaccordingtoprogressthroughthelearnedprocessofdi erentiation.Becausethepopulationmayactuallydi erentiateintomultipleseparatelineages,Monocleallowstheprocesstobranch,andcanassigneachcelltothecorrectsub-lineage.Itsubsequentlyidentiesgeneswhichdistinguishdi erentstates,andgenesthataredi erentiallyregulatedthroughtime.Finally,itperformsclusteringonallgenes,toclassifythemaccordingtokinetictrends.ThealgorithmisinspiredbyandandextendsoneproposedbyMagweneetaltotime-ordermicroarraysamples[2].Monocledi ersfrompreviousworkinthreeways.First,single-cellRNA-Seqdatadi erfrommicroarraymeasurementsinmanyways,andsoMonoclemusttakespecialcaretomodelthemappropriatelyatseveralstepsinthealgorithm.Secondly,theearlieralgorithmassumesthatsamplesprogressalongasingletrajectorythroughexpressionspace.However,duringcelldi erentiation,multiplelineagesmightarisefromasingleprogenitor.Monoclecanndtheselineagebranchesandcorrectlyplacecellsuponthem.Finally,Monoclealsoperformsdi erentialexpressionanalysisandclusteringontheorderedcellstohelpauseridentifykeyeventsinthebiologicalprocessofinterest.2Single-cellexpressiondatainMonocle Themonoclepackagetakesamatrixofexpressionvalues,whicharetypicallyforgenes(asopposedtosplicevariants),ascalculatedbyCuinks[3]oranothergeneexpressionestimationprogram.Monocleassumesthatgeneexpressionvaluesarelog-normallydistributed,asistypicalinRNA-Seqexperiments.Monocledoesnotnormalizetheseexpressionvaluestocontrolforlibrarysize,depthofsequencing,orothersourcesoftechnicalvariability-whicheverprogramthatyouusetocalculateexpressionvaluesshoulddothat.Monocleisnotmeanttobeusedwithrawcounts,anddoingsocouldproducenonsenseresults.2.1TheCellDataSetclassmonocleholdssinglecellexpressiondatainobjectsoftheCellDataSetclass.TheclassisderivedfromtheBioconductorExpressionSetclass,whichprovidesacommoninterfacefamiliartothosewhohaveanalyzedmicroarrayexperimentswithBioconductor.Theclassrequiresthreeinputles:1.exprs,anumericmatrixofexpressionvalues,whererowsaregenes,andcolumnsarecells2.phenoData,anAnnotatedDataFrameobject,whererowsarecells,andcolumnsarecellattributes(suchascelltype,culturecondition,daycaptured,etc.)3.featureData,anAnnotatedDataFrameobject,whererowsarefeatures(e.g.genes),andcolumnsaregeneattributes,suchasbiotype,gccontent,etc.TheexpressionvaluematrixmusthavethesamenumberofcolumnsasthephenoDatahasrows,anditmusthavethesamenumberofrowsasthefeatureDatadataframehasrows.RownamesofthephenoDataobjectshouldmatchthecolumnnamesoftheexpressionmatrix.RownamesofthefeatureDataobjectshouldmatchrownamesoftheexpressionmatrix.YoucancreateanewCellDataSetobjectasfollows: #notrunfpkm_matrix-read.table("fpkm_matrix.txt")sample_sheet-read.delim("cell_sample_sheet.txt")gene_ann-read.delim("gene_annotations.txt")pd-new("AnnotatedDataFrame",data=sample_sheet)fd-new("AnnotatedDataFrame",data=gene_ann)HSMM-new("CellDataSet",exprs=as.matrix(fpkm_matrix),phenoData=pd,featureData=fd)Itisoftenconvenienttoknowhowmanyexpressaparticulargene,orhowmanygenesareexpressedbyagivencell.Monocleprovidesasimplefunctiontocomputethosestatistics:Monocle:Di erentialexpressionandtime-seriesanalysisforsingle-cellRNA-SeqandqPCRexperiments ##T0_CT_A081472238NANANA##num_genes_expressedPseudotimeState##T0_CT_A0197707.2001##T0_CT_A0391802.7161##T0_CT_A0585282.2721##T0_CT_A0670966.4611##T0_CT_A0775903.4021##T0_CT_A08770220.3002Thisdatasethasalreadybeenlteredusingthefollowingcommands: valid_cells-row.names(subset(pData(HSMM),Cells.in.Well==1&Control==FALSE&Clump==FALSE&Debris==FALSE&Mapped.Fragmentsက1e+06))HSMM-HSMM[,valid_cells]Onceyou'veexcludedcellsthatdonotpassyourqualitycontrollters,youshouldverifythattheexpressionvaluesstoredinyourCellDataSetfollowadistributionthatisroughlylognormal: #Log-transformeachvalueintheexpressionmatrix.L-log(exprs(HSMM[expressed_genes,]))#Standardizeeachgene,sothattheyareallonthesamescale,Thenmelt#thedatawithplyrsowecanplotiteasily'melted_dens_df-melt(t(scale(t(L))))#Plotthedistributionofthestandardizedgeneexpressionvalues.qplot(value,geom="density",data=melted_dens_df)+stat_function(fun=dnorm,size=0.5,color="red")+xlab("Standardizedlog(FPKM)")+ylab("Density")##Warning:Removed2854443rowscontainingnon-finitevalues(stat_density). 4Basicdi erentialexpressionanalysis Di erentialgeneexpressionanalysisisacommontaskinRNA-Seqexperiments.Monoclecanhelpyoundgenesthataredi erentiallyexpressedbetweengroupsofcellsandassessesthestatisticalsigncanceofthosechanges.Thesecomparisonsrequirethatyouhaveawaytocollectyourcellsintotwoormoregroups.ThesegroupsaredenedbycolumnsinthephenoDatatableofeachCellDataSet.Monoclewillassessthesigncanceofeachgene'sexpressionlevelacrossthedi erentgroupsofcells.Performingdi erentialexpressionanalysisonallgenesinthehumangenomecantakeasubstantialamountoftime.Foradatasetaslargeasthemyoblastdatafrom[1],whichcontainsseveralhundredcells,theanalysiscantakeseveralhoursonasingleCPU.Let'sselectasmallsetofgenesthatweknowareimportantinmyogenesistodemonstrateMonocle'scapabilities: marker_genes-row.names(subset(fData(HSMM),gene_short_name%in%c("MEF2C","MEF2D","MYF5","ANPEP","PDGFRA","MYOG","TPM1","TPM2","MYH2","MYH3","NCAM1","TNNT1","TNNT2","TNNC1","CDK1","CDK2","CCNB1","CCNB2","CCND1","CCNA1","ID1")))Monocle:Di erentialexpressionandtime-seriesanalysisforsingle-cellRNA-SeqandqPCRexperiments plot_spanning_tree(HSMM) HSMM_filtered-HSMM[expressed_genes,pData(HSMM)$State!=3]my_genes-row.names(subset(fData(HSMM_filtered),gene_short_name%in%c("CDK1","MEF2C","MYH3")))cds_subset-HSMM_filtered[my_genes,]plot_genes_in_pseudotime(cds_subset,color_by="Time") Monocle:Di erentialexpressionandtime-seriesanalysisforsingle-cellRNA-SeqandqPCRexperiments