Best practice for mRNA microarray

Note : Please don't use it if you are not the fan of our biotrainee, Thanks.

Install required packages by the codes below:

source("http://bioconductor.org/biocLite.R") 
install.packages('devtools')
BiocInstaller::biocLite("jmzeng1314/biotrainee")
library(biotrainee)

But if you are in China, you should use the codes below:

options("repos" = c(CRAN="https://mirrors.tuna.tsinghua.edu.cn/CRAN/"))
install.packages("devtools",
			   repos="https://mirrors.tuna.tsinghua.edu.cn/CRAN/")
library(devtools) 
source("https://bioconductor.org/biocLite.R") 
options(BioC_mirror="https://mirrors.ustc.edu.cn/bioc/")  
BiocInstaller::biocLite(c('airway','DESeq2','edgeR','limma')) 
BiocInstaller::biocLite(c('ALL','CLL','pasilla','clusterProfiler')) 


library(devtools) 
source("https://bioconductor.org/biocLite.R") 
options("repos" = c(CRAN="https://mirrors.tuna.tsinghua.edu.cn/CRAN/"))
options(BioC_mirror="https://mirrors.ustc.edu.cn/bioc/")  
BiocInstaller::biocLite('org.Hs.eg.db')
install.packages("remotes",repos="https://mirror.lzu.edu.cn/CRAN/")
BiocInstaller::biocLite("jmzeng1314/biotrainee")
install.packages("pheatmap",repos="https://mirror.lzu.edu.cn/CRAN/")

It will install many other packages for you automately, such as : ALL, CLL, pasilla, airway ,limma，DESeq2，clusterProfiler , that's why it will take a long time to finish if all of these packages are not installed before in your computer.

Then run step1 :

It always not very easy to download data if you are in China, so I also upload the file GSE42872_raw_exprSet.Rdata , you can load it directly.

if(F){
  library(GEOquery)
  gset <- getGEO('GSE42872', destdir=".",
                 AnnotGPL = F,
                 getGPL = F)
  save(gset,'GSE42872.gset.Rdata')
}
load('GSE42872_eSet.Rdata')
b = eSet[[1]]
raw_exprSet=exprs(b) 
group_list=c(rep('control',3),rep('case',3))
save(raw_exprSet,group_list,
     file='GSE42872_raw_exprSet.Rdata')

Then step2:

Try to understand my codes, how did I filter the probes by the annotation of each microarry, and how I check the group information for the different samples in each experiment.

Including PCA and Cluster figures, as below:

Please ensure that you do run those codes by yourself !!!

Then step3:

Normally we will do differential expression analysis for the microarray, and LIMMA is one of the best method, so I just use it. If the expression matrix(raw counts ) comes from mRNA-seq, you can also choose DESeq based on negative binomial (NB) distributions or baySeq and EBSeq.

Once DEG finished, we can choose top N genes for heatmap as below: