Question: How to Set Up Contrasts for Case-Control Analysis in Limma While Adjusting for Covariates?
1
0
Entering edit mode
@31eec9dd
Last seen 21 days ago
United States

Hello Bioconductor community,

I am currently analyzing methylation array data using the EPIC array. The study is case-control, and my goal is to identify differentially methylated positions (DMPs) and differentially methylated regions (DMRs).

However, I'm unsure whether I need to set up a contrast for the case-control comparison while keeping the intercept at 0,

Or as the second group of codes if I should proceed using the limma package without specifying a contrast which is going to be the reference group.

My goal is to get DMPs with adjusting for age and gender.

The codes are below


S_Group <- factor(sample_sheet$HTN) # Case and Control

age <- sample_sheet$AGE # Continuous variable
gender <- factor(sample_sheet$GENDER) # Male and Female


# 1A. Adjusting for Age and gender 

# Design Matrix
design <- model.matrix(~0 + S_Group + age + gender , data=sample_sheet)


colnames(design)

# Naming the columns
colnames(design) <- gsub("genderM", "sexMale", colnames(design))

# Fitting the Linear Model

fit <- lmFit(mVals, design)

str(design)

# Contrast Matrix (Case vs Control)
contMatrix <- makeContrasts(CaseVsControl = S_Group1 - S_Group2, levels = design)

contMatrix
# Applying the Contrast
fit2 <- contrasts.fit(fit, contMatrix)
fit2 <- eBayes(fit2)


# Match the rownames of m_vals to the IlmnID column of annEPICv2 to extract relevant annotation
annEPICv2Sub <- annEPICv2[match(rownames(mVals), annEPICv2$IlmnID),]

# Extract top differentially methylated positions (DMPs)
DMPs1 <- topTable(fit2, num=Inf, coef=1, genelist=annEPICv2Sub)
head(DMPs1)

Yesterday, I realized that the approach I was using with the limma package was performing a pairwise comparison. To address this, I used the following code:


design <- model.matrix(~ S_Group + age + gender , data=sample_sheet)
fit <- lmFit(mVals, design)
fit <- eBayes(fit)
topTable(fit)

Thank you

MethylationArray EpigeneticsWorkflow limma Regression • 253 views
ADD COMMENT
2
Entering edit mode
@james-w-macdonald-5106
Last seen 1 day ago
United States

I don't know what you mean by 'a pairwise comparison', but the two analyses are equivalent, except in the second analysis you want to do topTable(fit, 2), because the comparison between the two groups is coefficient 2. If you don't specify the contrast, you will get an F-test for all the coefficients instead, which is uninteresting.

Login before adding your answer.

Traffic: 364 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6