paper.3
2018-12-19 本文已影响0人
Mabel娜
paper.3
workflow.JPG
Identifying miRNA/mRNA negative regulation pairs in colorectal cancer
这篇是曾大大老早之前推荐的,再来一篇,快乐快乐
看起来简洁明了,直击痛点哎
背景不表,来学绝招
workflow.JPG
1. data
TCGA分别获取miRNAseq,mRNAseq数据, 共261个样本,包括253 CRC samples,8 normal tissue samples
2. screening of differential genes and miRNA
2种方法取交集,包括:
Samr package(delta=1, FC>2, FDR<5%);
limma package ( p.adj<.05, FC>2);
3. target genes of the differential miRNA
首先通过miRWalk2.0查找diff-miRNA的target mRNAs;
然后2者pearson rank correlation;
最后筛选出差异表达并负相关的基因;
4. determination of the disease-related miRNA/mRNA
终于构建好了对子,接下来就是GO/KEGG analysis;
5. construction of PPI network
首先对disease-related miRNA/mRNA进行PCA analysis;
接着对变化最大的disease-related miRNA进行生存分析;
最后把对子扔进STRING,构建PPI network;
6. network analysis
network的topology parameters and regulation networks。
这样看来,ROC,AUC,机器学习的分类算法是必须要填坑的啊