既然你问这种问题，我觉得tcga大文章肯定不适合你看，真心有点难度来。
看bmc,SR,oncotarget,plos one 里面的套路文章吧：
比较多的是下面一些套路
一是直接找全局miRNA/mRNA表达对
GSE35982,BMC,2012
TCGA,BMC,2012
(GSE11016,GSE12105,GSE16441,GSE23085),BMC,2013
TCGA,Nat Commun. 2015
TCGA,PNAS,2013
GSE19783 ,Tumor Biology,2015
TCGA,PLOS ONE,2017
- Prediction of clinical outcome in glioblastoma using a biologically relevant nine-microRNA signature
tcga,Molecular Oncology,2015
- Genomic and Epigenomic Landscapes of Adult De Novo Acute Myeloid Leukemia
- Meta-analysis of microRNA expression in lung cancer
GEO,International journal of cancer 2012
二是仅仅是针对感兴趣基因来分析公共数据
SAMSN1 Is Highly Expressed and Associated with a Poor Survival in Glioblastoma Multiforme(2013 PLOS ONE)
- Cancer-Testis Antigen Expression in Serous Endometrial Cancer with Loss of X Chromosome Inactivation]http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0137476)
TCGA,PLOS ONE,2014
TCGA,PLOS ONE,2014
(GSE 4271, GSE4412),PLOS ONE,2014
TCGA,PLOS ONE,2014
TCGA,PLOS ONE,2016
TCGA,Nucl Acids Res (2015)
TCGA,PLOS ONE,2016
首先指明一个现象，多种癌症里面，尤其是HNSCC中，STAT3基因的第705个酪氨酸(Y,tyrosine)的磷酸化导致它被过度激活。
所以，就针对STAT3基因抓取了HNSCC里面的各种关于它的数据：Mutation, mRNA expression, promoter methylation, and copy number alteration data were extracted from TCGA and examined in the context of pSTAT3(Y705) protein expression.从表达量的相关性找到了1279 相关genes
TCGA，2015 PLOS ONE
从题目就知道他们做了什么分析：In this study, the miRNA profiles in 327 HCC patients, including 327 tumor and 43 adjacent non-tumor tissues, from The Cancer Genome Atlas (TCGA) Liver hepatocellular carcinoma (LIHC) were analyzed.里面找到了差异的miRNA(Differentially expressed miRNAs (DEmiRNA) )，然后用了miRNA pathway analysis工具miRPath做了功能分析而已。当然，生存分析也必不可少啦。
TCGA,2012 PLOS ONE
这个稍微有点不同，作者用自己的2009年发表的iCluster工具来灌水，下载了TCGA glioblastoma (GBM)的3种数据，iCluster was applied using 1,599 copy number features, 1,515 DNA methylation features, and 1,740 expression features。用iCluster把GBM分成了3个亚型。还跟前人单纯的用表达数据，或者甲基化数据的分类做了比较，还跟PCA比较，必须是自己的好呀
TCGA,2016 PLOS ONE
用的是公共数据，用了多种数据来把样本分类，自己开发了一个简单粗糙的分类方法而已。
三是补充一点简单的实验，用TCGA数据做佐证而已。
