GEOquery simple example for four cancer/normal samplesets

bio361-geoquery-walkthrough.R

## ----message=FALSE,warning=FALSE----------------------------------------------
pkgs = c(
  "ggplot2",
  "GEOquery",
  "SummarizedExperiment"
)
ins = installed.packages(repos = BiocManager::repositories())
for(pkg in pkgs) {
  if(!(pkg %in% rownames(ins)))
    BiocManager::install(pkg)
}


## -----------------------------------------------------------------------------
## GEOquery
## -----------------------------------------------------------------------------
library(GEOquery)
options("download.file.method.GEOquery"="curl")
gse = getGEO("GSE103512")[[1]]


## -----------------------------------------------------------------------------
library(SummarizedExperiment)
se = as(gse, "SummarizedExperiment")


## ----geoquery20,echo=TRUE,cache=TRUE------------------------------------------
with(colData(se),table(`cancer.type.ch1`,`normal.ch1`))


## ----sds,cache=TRUE,echo=TRUE-------------------------------------------------
sds = apply(assay(se, 'exprs'),1,sd)
dat = assay(se, 'exprs')[order(sds,decreasing = TRUE)[1:500],]


## ----mds,echo=TRUE,cache=TRUE-------------------------------------------------
mdsvals = cmdscale(dist(t(dat)))
mdsvals = as.data.frame(mdsvals)
mdsvals$Type=factor(colData(se)[,'cancer.type.ch1'])
mdsvals$Normal = factor(colData(se)[,'normal.ch1'])


## ----mdsplot,echo=TRUE,fig.align='center'-------------------------------------
library(ggplot2)
ggplot(mdsvals, aes(x=V1,y=V2,shape=Normal,color=Type)) + 
    geom_point( alpha=0.6) + theme(text=element_text(size = 18))