Tools for the analysis of high- dimensional si.pdfVIP免费

Inasingleorganism,mostcellshavethesamegenome,butspecificgeneexpressionvariesacrossdifferenttis­suesandcelltypes.Anygiventissueorcelltypeexpres­ses~11,000–13,000genes,ofwhich~3,000–5,000haveacell­type­specificexpressionpattern,whereastheremaininggenesareubiquitouslyexpressed1.Theseuniquepatternsofgeneexpressiontranslatetodiffer­encesattheproteinlevelbetweendifferentcelltypesandresultinthevastarrayofcellularphenotypesfoundthroughoutthebody.Therefore,asnapshotofthegeneexpressionprofileofacellcanbeindicativeofitspheno­type.OwingtothelimitedamountofRNApresentineachcell,geneexpressionprofilingwashistoricallyperformedonpooledcells,butthisbulksequencingapproachobscuredthepotentialcellheterogeneityinasampleortissue2.Forexample,inapoolofdevelop­ingprogenitorcells,differentcellsmightbeprimedtomakedistinctfatedecisionsbutthesetranscriptionalprogrammesareindistinguishableinabulkanalysisoftheaveragegeneexpressionintheprogenitorpool.Thedevelopmentoftechnologiesthatcanisolatethousandstotensofthousandsofcellsandassesstheirgeneexpressionprofilesatthesingle­celllevelhasena­bledresearcherstodissectthiscellularheterogeneityandworktowardsabetterunderstandingofphysiology,biologicaldevelopmentanddisease2–6.Forexample,researchersgeneratedanimprovedquantitativemapofthecelltypespresentinthedevelopinghumankidney,whichhasprovidedinsightsintorenalphysiology7.Anothersingle­cellstudydemonstratedthesimilaritiesbetweenfetalhumankidneyandhumankidneyorga­noids,reaffirmingtheutilityofkidneyorganoidsasamodelforthestudyofdiseaseandfordrugscreening8.However,derivingbiologicalinsightsfromsingle­cellRNAsequencing(scRNA­seq)methodsdemandsthatresearchershandlethelargevolumeofdatageneratedbythesetechnologiesandtheiraccompanyingsourcesoftechnicalnoise9.Addressingthescaleandcomplexityofthesedatasetsthusrequiresacomplexecosystemofcomputationalmethods.BeyondscRNA­seqanalysis,otheravailabletechno­logiescanprofilegenomes10,methylationpatterns11andchromatinaccessibilitypatterns12,13atthesingle­c...