Difference between revisions of "Using Bioconductor To Analyse Microarray Data"

Revision as of 01:19, 27 July 2009

Software Requirements

• R, get from [CRAN]
• Bioconductor, get from [Bioconductor]
• Bioconductor packages. Install as needed:
  • Biobase
  • GEOquery - [1]
  • Limma

source("http://www.bioconductor.org/biocLite.R")
biocLite("PACKAGE")

Obtaining GEO Datasets

• Open a R terminal
• Load Biobase and GEOquery packages

libary(Biobase)
library(GEOquery)

• Can load:
  • datasets - GDS
  • measurements - GSM
  • platforms - GPL
  • series - GSE

gds <- getGEO("GDS2946")  #load GDS162 dataset
Meta(gds)  #show extracted meta data
table(gds)[1:10,]  #show first ten rows of dataset
eset <- GDS2eSet(gds, do.log=TRUE)  #convert to expression set, by default obtains annotation (GPL) data with log2 transformation
pData(eset)  #phenotype data
sampleNames(eset)  #sample names (GSM)

• see [Peter Cock's Page] or [GEOquery Documentation] for more information.

Microarray Analysis

• set up design matrix. Use a different integer for each treatment group. The following example is for a contrast between the first seven groups and the last eight groups. For details on other design matrices see chapter 8 of [limma User Guide]

library(limma)  #load limma package
library(affyPLM)  #load affyPLM package
eset.norm <- normalize.ExpressionSet.quantiles(eset)  #normalize expression set by quantile method
pData(eset)  #to see phenotype annotation data
design <- model.matrix(~(c(1,1,1,1,1,1,1,0,0,0,0,0,0,0,0)),eset)  #set design matirx
colnames(design) <- c("resistant","sensitive")  # give names to the treatment groups
design  #check the design matrix
fit <- lmFit(eset.norm,design)  #Fit data to linear model
fit.norm.eb <- eBayes(fit.norm)  #Empirical Bayes
write.csv(fit.norm.eb, file="filename.csv")  #write to CSV file