Question: Limma voom mechanism
gravatar for icxa0409
3 months ago by
United States
icxa04090 wrote:

Is it possible to transform RNA Seq data present as a txt file containing gene, log2foldchange, baseMean, stderror, pvale, padjusted etc. using voom? If yes, then please can someone tell me how to do that?



ADD COMMENTlink modified 3 months ago by Aaron Lun15k • written 3 months ago by icxa04090
gravatar for Aaron Lun
3 months ago by
Aaron Lun15k
Cambridge, United Kingdom
Aaron Lun15k wrote:

Yes. Read Sections 15 and 18 of the limma user's guide.

Edit: There seems to be some confusion in the wording of your question, now that I look at it more closely. When we talk about RNA-seq data for DE analyses, we are usually referring to a matrix of read counts for genes (rows) and samples (columns). If you have this, then you can use voom as described in the limma user's guide. However, It is not correct to describe a text file of log-fold changes, p-values, etc. as "RNA-seq data". Those values represent the results of an existing DE analysis - probably DESeq2, judging by the column names - and you can't go from the analysis results back to the raw data. If you want to use voom, get the count matrix.

ADD COMMENTlink modified 3 months ago • written 3 months ago by Aaron Lun15k

Thank you for your help. I have read them. But still I am unable to figure which commands/functions to use for my task. 

ADD REPLYlink written 3 months ago by icxa04090

I'm not sure if your reply landed before or after Aaron's edit, so I'll just provide a little more help.

The text file that it sounds like you want to make can be had by saving the result of a final call to "topTable" after the successful analysis of your RNA-seq data.

It's the "successful analysis" bit that's the tricky part, however, and the limmaUserGuide shows a ton of examples that result in a "successful analysis" of expression data to identify fold changes and pvalues from experimental designs of many flavors.

In broadest of strokes, your steps would look something like:

y <- DGEList(counts, group=<something>)
yf <- ## filter y to remove low counts
design <- model.matrix(~ group)
vm <- voom(yf, design)
fit <- lmFit(vm, vm$design)
fit <- eBayes(fit)
res <- topTable(fit, 'coef', n=Inf)
write.csv(res, 'result.txt')

You have to get the above part right, as it can differ in many ways. Again, the limmaUserGuide and edgeR guide show you how to do that in a variety of ways, but you'll have to take the time to read those guides, and the even longer amount of time required to understand it ;-)

Once you do, though, "res" from is above is the one you want to save and explore for your results.

ADD REPLYlink written 3 months ago by Steve Lianoglou12k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 169 users visited in the last hour