We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
DLGCN :基于图卷积网络的药物-lncRNA 关联预测.
- Authors
朱济村; 周旭; 侯斐; 曹新玉; 姜伟瘁
- Abstract
To realize high-LhroughpuL idcnLificaLion of new drug-lncRMA associaLions, we propose a new meLhod DLGCN (Drug-LneRMA graph convolLiLion noLwork) Lo idenLify poLenLial drug-lncRMA associaLions・ Fit'sL, we consLrucL drug-drug and lncRM A-Inc RM A similat'iLy ncL works based on drug sLrucLurc information and lncRMA sequence informaLion, and Lhen combine Lhem wiLh known drug-lncRMA associaLions Lo consLrucL drug-lncRMA hcLcrogcncous neLwork・ MexL, Lhe aLLenLion mechanism and graph convoluLion opcraLion arc applied Lo Lhe noLwork Lo learn Lhe low dimensional fcaLurcs of drugs and Inc RM As. The new drug-lncRMA associaLions arc prcdicLcd based on Lhe inLegraLed low dimensional fcaLurcs ・ DLGCN idenLified Lhe drug-lncRMA associaLions wiLh an AU ROC (Area under Lhe receiver opcraLor charac Leris Lie) of 0.843 1, which is superior Lo classical machine learning; meLhods and common deep learning meLhods ・ In addiLion, DLGCN predieL LhaL curcumin could rcgidaLc M AL ATI, which has been confirmed by rcccnL sLudics. DLGCN can effeeLively predieL drug-lncRNA associaLions, which provides an imporLanL rcfcrcncc for idcnLincaLion of new Lumor LhcrapcuLic Lat'gcLs and dev elopmen L of anLi-canccr drugs.
- Publication
Chinese Journal of Bioinformatics, 2024, Vol 22, Issue 2, p93
- ISSN
1672-5565
- Publication type
Article
- DOI
10.12113/202212004