Question: Differential Expression Analysis on miRNA Normalized data
0
asalimih30 wrote:

Hello, I have a normalized miRNA data matrix of 6 groups. every group has 3 samples.
density plot of each sample is: I have no access to the raw count data. So for analysing this matrix I used limma package. But results I get are not as it must be (based on the corresponding paper).
my R code:

``````gr <- factor(rep(c("DC1","DC2","DC3","DC4","DC5","DC6"),each=3))
miR\$samples\$group <- gr;
design <- model.matrix(~0+gr)
colnames(design) <- gsub("gr", "", colnames(design))

contr.matrix <- makeContrasts(
DC1.DC2 = DC2-DC1,
DC2.DC3 = DC3-DC2,
DC3.DC4 = DC4-DC3,
DC4.DC5 = DC5-DC4,
DC5.DC6 = DC6-DC5,
levels = colnames(design))

miR.v <- voom(miR, design, plot=TRUE)
miR.v.fit <- lmFit(miR.v, design)
miR.v.fit <- contrasts.fit(miR.v.fit, contrasts=contr.matrix)

miR.e.fit <- eBayes(miR.v.fit)
dt.e <- decideTests(miR.e.fit,lfc = 1)
summary(dt.e)
``````

the voom's plot: output of summary(dt.e) function:

``````> summary(dt.e)
DC1.DC2 DC2.DC3 DC3.DC4 DC4.DC5 DC5.DC6
Down         4       0       1       1       4
NotSig    1970    1983    1975    1982    1978
Up           9       0       7       0       1
``````

As you can see in the code, i want to know upregulation and downregulation of miRNAs between every two consecutive groups.

so here are my questions:

1. Is the matrix really normalized?(how can i know)
2. if it is normalized can i use limma voom function on normalized data? if not what should i do?
3. Am i doing anything wrong?

every help would be appreciated so much