PeakidentificationforChIP-seqdatawithnocontrols

Chromatinimmunoprecipitationfollowedbysequencing(ChIP-seq)isincreasinglybeingusedforgenome-wideprofilingoftranscriptionalregulation,asthistechniqueenablesdissectionofthegeneregulatorynetworks.Withinputascontrol,avarietyofstatisticalmethodshavebeenproposedforidentifyingtheenrichedregionsinthegenome,i.e.,thetranscriptionalfactorbindingsitesandchromatinmodifications.However,whentherearenocontrols,whetherpeakcallingisstillreliableawaitssystematicevaluations.Toaddressthisquestion,weusedaBayesianframeworkapproachtoshowtheeffectivenessofpeakcallingwithoutcontrols(PCWC).UsingseveraldifferenttypesofChIP-seqdata,wedemonstratedtherelativelyhighaccuracyofPCWCwithlessthana5%falsediscoveryrate(FDR).Comparedwithpreviouslypublishedmethods,e.g.,themodel-basedanalysisofChIP-seq(MACS),PCWCisreliablewithlowerFDR.Furthermore,tointerpretthebiologicalsignificanceofthecalledpeaks,incombinationwithmicroarraygeneexpressiondata,geneontologyannotationandsubsequentmotifdiscovery,ourresultsindicatePCWCpossessesahighefficiency.Additionally,usinginsilicodata,onlyasmallnumberofpeakswereidentified,suggestingthesignificantlylowFDRforPCWC.