Log-cpm values from limma
1
0
Entering edit mode
@14ef1b09
Last seen 3 days ago
Egypt

I have a performed a typical differential expression analysis using limma voom and I want to extract the log cpm values to draw ROC curves. But I want to extract the log cpm values produced specifically by limma because they are normalised and free of batch effect that was added to the design matrix. Is there a function to do that?

Thanks in advance

limma edgeR RNASeq • 317 views
ADD COMMENT
3
Entering edit mode
@gordon-smyth
Last seen 6 hours ago
WEHI, Melbourne, Australia

The log-cpm values from voom() are not free of batch effect. If you want batch-corrected log-cpm values, we recommend cpm() followed by removeBatchEffect().

ADD COMMENT
0
Entering edit mode

I have performed RUVg correction and extracted W, factors of unwanted variation. According to limma guide,

In the limma approach to RNA-seq, read counts are converted to log2-counts-per-million (logCPM) and the mean-variance relationship is modeled either with precision weights or with an empirical Bayes prior trend.

design <- model.matrix(~0+dge$samples$Stage + W)

Does this mean if I extract the logcpm counts using

   v=voom(dge, design = design, plot = TRUE)
   lcpm=v$E 

the extracted counts will be free of batch effect and unwanted variation?v

ADD REPLY
2
Entering edit mode

No, it does not mean that. voom() uses the design matrix with the W covariates, to compute precision weights but not to adjust the log-cpm values. Adjustment for batch factors, such as the W matrix from RUV, is done during the differential expression analysis. At no stage are the original counts or log-cpm values adjusted.

I gave you the correct answer three weeks ago. If you want log-cpm values that are batch adjusted, you need to use cpm and removeBatchEffect.

ADD REPLY

Login before adding your answer.

Traffic: 730 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