Question: Limma paired sample analysis
1
2.2 years ago by
anandprem179210 wrote:

I have an Affymetrix single channel DNA microarray dataset where normal and diseased tissue are taken from the same organ from 5 persons. So I have 10 CEL files with me. Now I want to analyse the differential gene expression between the normal and diseased tissues. I think I should use Limma Paired samples design matrix and compute paired moderated t test. Limma user guide. 9.4.1 pg no: 42 and 43.

Can anyone please confirm whether my approach is correct or Should I follow different approach ?

My phenodata file looks like this

 FileName Subject Tissue 100.CEL 1 NORMAL 101.CEL 1 DISEASED 102.CEL 2 NORMAL 103.CEL 2 DISEASED 104.CEL 3 NORMAL 105.CEL 3 DISEASED 106.CEL 4 NORMAL 107.CEL 4 DISEASED 108.CEL 5 NORMAL 109.CEL 5 DISEASED

My code

library(affy)

library(limma)

# Read all CEL FIles and put into an affybatch

#Importing the phenotype data

#Visualise the phenotype data
pData(affy.data)

# Normalize the data
eset = rma(affy.data)

# DIFFERENTALLY GENE EXPRESSION ANALYSIS

library(limma)
pData(eset)
Subject <- factor(eset$Subject) Tissue <- factor(eset$Tissue, levels = c("Normal", "Diseased"))

design <- model.matrix(~Subject+Tissue)
design

fit <- lmFit(eset, design)
eBayesfit <- eBayes(fit)

View(eBayesfit)
topTable(eBayesfit, coef="TissueDiseased")

modified 2.2 years ago • written 2.2 years ago by anandprem179210
1
2.2 years ago by
Aaron Lun25k
Cambridge, United Kingdom
Aaron Lun25k wrote:

Looks fine to me. You should probably do some filtering on abundance, though. See the relevant section of the user's guide and case studies for examples.

P.S. limma should be a separate tag, otherwise the maintainers don't get notified.