欢迎来到天天文库
浏览记录
ID:41101320
大小:584.86 KB
页数:22页
时间:2019-08-16
《Algorithm to Identify Frequent Coupled Modules(0001)》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、AlgorithmtoIdentifyFrequentCoupledModulesfromTwo-LayeredNetworkSeriesPresentation:YangLiBackground:Background:�Currentnetworkanalysismethodsallfocusononeormultiplenetworksofthesametype.�Cellsareorganizedbymulti-layernetworks.�Differentkindsofnetworksinfluenceeachother.�RNA-se
2、qhasprovidedabundantmulti-levelprofilingdata------expressiondata&splicingdata.�Splicingcoupleswithtranscriptionwiththefollowingthreesplicingreactions------capping,splicing,cleavageandpolyadenylation.CouplednetworkCouplednetworkEssentialsEssentials:•RNA-seqdatasetsselection,pr
3、ocessingandnetworkconstruction•Non-uniformsamplingforfastcomputation•Formulatetheproblemasatenser-based0-1nonlinearintegerprogrammingmodel•Relaxthe0-1nonlinearintegerprogrammingintoacontinuouscomputationalproblem•OptimizationalgorithmandpatternextractionRNA-seqdatasetsselecti
4、on:RNA-seqdatasetsselection:�Database:NCBI’sSequenceReadArchive(SRA)�Criterions:allhumanRNA-seqdatasets,eachofwhichcontainsatleastsixsamples(38datasets)---------Theminimumvalueforrobustcorrelationestimationis6.Dataprocessing:Dataprocessing:①Foreachdataset,weusedtheTophattoolt
5、omapshortreadstothehg18referencegenome(reportonlytheoptimalalignmentanddiscardthosereadsthatmapequallywelltomultiplepositions).②WeappliedthetranscriptassemblytoolCufflinkstoestimateexpressionsforalltranscriptswithknownUCSCtranscriptionannotations.③Calculatedtheinclusionrateof
6、eachexonineverysample.Networkconstruction:Networkconstruction:�ForeachRNA-seqdatasetaweightedgeneco-expressionnetworkcanbeconstructed(edgesweightsarecorrelationsbetweentheexpressionprofilesoftwogenes).�Thesameprocedureisappliedtobuildaweightedexonco-splicingnetworkforthesameR
7、NA-seqdataset(edgeweightsrepresentcorrelationsbetweentheinclusionratesoftwoexons).Weightscomputation:Weightscomputation:①Computeleave-one-outPearsoncorrelationefficientr;②ComputeFisher'stransformationscoren−3⎛1+r⎞;z=ln⎜⎟2⎝1−r⎠①Standardizethez-scorestoenforcezeromeanandunitvar
8、iance;②Setvirtualsamplesizen'=10,andcompute⎛2⎞exp⎜z⎟−1⎝n'−3⎠.r'=⎛2⎞e
此文档下载收益归作者所有