Integrative modeling of multi-omics datato identify cancer drivers and inferpatient-specific gene activity多组学数据的集成建模 识别癌症驱动力和推断 患者特异性基因活性

《Integrative modeling of multi-omics datato identify cancer drivers and inferpatient-specific gene activity多组学数据的集成建模 识别癌症驱动力和推断 患者特异性基因活性》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、Paveletal.BMCSystemsBiology(2016)10:16DOI10.1186/s12918-016-0260-9METHODOLOGYARTICLEOpenAccessIntegrativemodelingofmulti-omicsdatatoidentifycancerdriversandinferpatient-specificgeneactivityAnaB.Pavel1,2*,DmitriySonkin3andAnupamaReddy4AbstractBackground:Highthroughputtechnologieshavebeenusedtoprofile

2、genesinmultipledifferentdimensions,suchasgeneticvariation,copynumber,geneandproteinexpression,epigenetics,metabolomics.Computationalanalysesoftentreatthesedifferentdatatypesasindependent,leadingtoanexplosioninthenumberoffeaturesmakingstudiesunder-poweredandmoreimportantlydonotprovideacomprehensivevi

3、ewofthegene’sstate.Wesoughttoinfergeneactivitybyintegratingdifferentdimensionsusingbiologicalknowledgeofoncogenesandtumorsuppressors.Results:Thispaperproposesanintegrativemodelofoncogeneandtumorsuppressoractivityincellswhichisusedtoidentifycancerdriversandcomputepatient-specificgeneactivityscores.We

4、havedevelopedaFuzzyLogicModeling(FLM)frameworktoincorporatebiologicalknowledgewithmulti-omicsdatasuchassomaticmutation,geneexpressionandcopynumbermeasurements.Theadvantageofusingafuzzylogicapproachistoabstractmeaningfulbiologicalrulesfromlow-levelnumericaldata.Biologicalknowledgeisoftenqualitative,t

5、huscombiningitwithquantitativenumericalmeasurementsmayleveragenewbiologicalinsightsaboutagene’sstate.Weshowthattheoncogenicandalteredtumorsuppressingstateofagenecanbebettercharacterizedbyintegratingdifferentmolecularmeasurementswithbiologicalknowledgethanbyeachdatatypealone.Wevalidatethegeneactivity

6、scoreusingdatafromtheCancerCellLineEncyclopediaanddrugsensitivitydataforfivecompounds:BYL719(PIK3CAinhibitor),PLX4720(BRAFinhibitor),AZD6244(MEKinhibitor),Erlotinib(EGFRinhibitor),andNutlin-3(MDM2inhibitor).Theintegrativescoreimprovespredictionofdrugsensitivityfortheknowndrugtargetsofthesecompoundsc

7、omparedtoeachdatatypealone.Thegeneactivityscoresarealsousedtoclustercolorectalcancercelllines.TwosubtypesofCRCswerefoundandpotentialcancerdriversandtherapeutictargetsforeachofthesubtypeswereidentified