开放期刊系统

基于Transformer特征学习与多源异构数据融合的泛癌驱动模块识别

邓 诗宇

摘要

针对传统方法处理高维稀疏数据能力不足的问题,本文提出基于Transformer的深度特征学习框架。该框架以TCGA泛癌数据为基础,融合基因组、转录组及蛋白质互作网络等多源异构数据,利用多头自注意力机制捕捉基因间的非线性与长距离依赖关系,实现了对潜在驱动基因的深度语义表征,为多组学数据融合与癌症分子机制解析提供了新思路。

关键词

癌症驱动模块;Transformer;多源异构数据;泛癌

全文:

PDF


参考

[1]DasJ,YuH.HINT:High-qualityproteininteractomesandtheirapplicationsinunderstandinghumandisease[J].BMCsystemsbiology,2012,6(1):1-12.[2]LeisersonMDM,VandinF,WuHT,etal.Pan-cancernetworkanalysisidentifiescombinationsofraresomaticmutationsacrosspathwaysandproteincomplexes[J].Naturegenetics,2015,47(2):106-114.[3]YuH,TardivoL,TamS,etal.Next-generationsequencingtogenerateinteractomedatasets[J].Naturemethods,2011,8(6):478-480.[4]TokarT,PastrelloC,RossosAEM,etal.mirDIP4.1—integrativedatabaseofhumanmicroRNAtargetpredictions[J].Nucleicacidsresearch,2018,46(D1):D360-D370.[5]VaswaniA,ShazeerN,ParmarN,etal.Attentionisallyouneed[J].Advancesinneuralinformationprocessingsystems,2017,30:5998-6008.[6]McLachlanGJ.Mahalanobisdistance[J].Resonance,1999,4(6):20-26.[7]Kanehisa,Minoru,andSusumuGoto.KEGG:kyotoencyclopediaofgenesandgenomes[J].Nucleicacidsresearch,2000,28(1):27-30.[8]JohnGTate,SallyBamford,etal.COSMIC:theCatalogueOfSomaticMutationsInCancer.[J].NucleicAcidsResearch,2019,47(D1):D941-D947.[9]DresslerL,BortolomeazziM,KeddarMR,etal.Comparativeassessmentofgenesdrivingcancerandsomaticevolutioninnon-cancertissues:anupdateoftheNetworkofCancerGenes(NCG)resource[J].GenomeBiol,2022,23(1):35.


(0 摘要 Views, 0 PDF Downloads)

Refbacks

  • 当前没有refback。