摘要
Inviewofpotentialapplicationtobiomedicaldiagnosis,tighttranscriptomedataqualitycontroliscompulsory.Usually,qualitycontrolisachievedusinglabelingandhybridizationcontrolsaddedatdifferentstagesthroughouttheprocessingofthebiologicRNAsamples.Thesecontrolmeasures,however,onlyreflecttheperformanceoftheindividualtechnicalmanipulationsduringtheentireprocessandhavenobearingastothecontinuedintegrityoftheRNAsampleitself.Herewedemonstratethatintrinsicstatisticalpropertiesoftheresultingtranscriptomedatasignalandsignal-variancedistributionsandtheirinvariancecanbeidentifiedindependentlyoftheanimalspeciesstudiedandthelabelingprotocolused.Fromtheseinvariantpropertieswehavedevelopedadatamodel,theparametersofwhichcanbeestimatedfromindividualexperimentsandusedtocomputerelativequalitymeasuresbasedonsimilaritywithlargereferencedatasets.Thesequalitymeasuresaddsupplementary,non-redundantinformationtostandardqualitycontrolestimatesbasedonspike-inandhybridizationcontrols,andareexploitableindataanalysis.AsoftwareapplicationforanalyzingdatasetsaswellasareferencedatasetforAB1700arraysareprovided.TheyshouldallowAB1700userstoeasilyintegratethismethodintotheiranalysispipeline,andmightinstigatesimilardevelopmentsforothertranscriptomeplatforms.
出版日期
2010年01月11日(中国期刊网平台首次上网日期,不代表论文的发表时间)