logCPM error in RStudio.

1

Entering edit mode

4.2 years ago

far.zi ▴ 10

Hi, I've beed doing RNA-Seq analysis using R. However, when I get the final edgeR_result file, the logCPM values are very high which is not true for the genes I'm studying since they are extremely low expressed. I've checked my count table many times and no mistake there. Can someone help me please to figure it out why I'm getting those high values? Thanks I also put the script I use in R in case needed:

cts <- read.csv("normCounts.txt", sep = '\t', row.names = 1, header = FALSE) 
group <- (c(1,1,1,1,2,2,2,2))
cts <- DGEList(cts, group)
#cts <- calcNormFactors(y)
design <- model.matrix(~group)
cts <- estimateDisp(cts, design)
fit <- glmQLFit(cts, design)
qlf <- glmQLFTest(fit, coef = 2)
write.table(qlf$table, file = "edgeR_result.txt")
res <- qlf$table

ggplot(res, aes(x=logCPM, y=logFC, color=PValue < 0.05)) + 
    geom_point() + geom_rug(sides = "l") + theme_minimal()

fit2 <- glmFit(cts, design)

enter image description here

R RNA-Seq software error • 1.2k views

ADD COMMENT • link updated 4.2 years ago by ATpoint 85k • written 4.2 years ago by far.zi ▴ 10

1

Entering edit mode

Are you feeding normalized counts into edgeR? Is this correct that you #-ed the calcNormFactors? Please add the plot you created.

ADD REPLY • link 4.2 years ago by ATpoint 85k

0

Entering edit mode

Yes, I am feeding the normalized values.

ADD REPLY • link 4.2 years ago by far.zi ▴ 10

0

Entering edit mode

That is not correct. Please read the manual. You must start from raw counts. There are also very few points in this plot. How many in total?

ADD REPLY • link 4.2 years ago by ATpoint 85k

0

Entering edit mode

OMG, I thought I need the norm counts. The total is 478.

ADD REPLY • link 4.2 years ago by far.zi ▴ 10

0

Entering edit mode

That is a very low gene number, so this is no normal RNA-seq, right? Yes, raw counts, I suggest you go through the edgeR manual first before continuing.