同journal同期同国人发表的一起拿过来学习啦,这个题目还短一些哎,意在研究这个RNA调控对子怎么做,好,这个直接把ceRNA机制拿出来啦
略过背景,直达技能
1. data resources and pretreatment
RNA-seq(coding and non-coding) data 和临床信息来自TCGA, 注释来自Ensembl;
miRNA expression profiles 来自UCSC Xena,分析成熟而非前体;
为了排除低表达基因,选取75%样本中lncRNA,mRNA raw count value >10, miRNA log2(RPM+1)>1 ;
2. differential analysis
HCC中癌组织和癌旁正常组织的lncRNA, mRNA用edgeR package;
miRNA用limma package;
abs(log(FC))>1 and p.val<.05;
3. survival analysis and gene network construction
只选择总生存期>90 days的病患,首先,采用单变量COX回归分析RNA与总生存期的相关性,候选biomarkers的p.val<.001, 然后采用多变量COX回归移除预后的非独立RNA,最后,通过多变量COX回归建立RNA评分矩阵,p.val<.01;
SurvivalROC package 允许事件依赖的受试者工作特性曲线ROC评估,并得到AUC;
4. CeRNA network construction
预后相关的RNA用于构建ceRNA network, 具体数据库为miRcode, TargetScan, miRTarBase, miRDB,(这里可以深挖就是为什么miRNA-mRNA pairs interactions要用3个数据库),联合2个对子构建ceRNA,并丢进cytoscape;
5. functional analysis
lncRNA-miRNA-mRNA构成的ceRNA network进行GO,KEGG enrichment analysis,clusterProfiler and GOplot package were used;
6. statistical processing
independent sample t-tests;
K-M analysis;
log-rank test.
图多又好看
网友评论