Limma voom with missing values from mass spectrometry

Question

[harrydolan.dc]

Hi everyone. Firstly thank you in advance for any help you can give, I am new to working with mass spec data and biostars has been immensely helpful.

I data from three technical repeats and three biological repeats (9 Controls, 9 Treated - 18 Total) Each time the mass spec was performed some genes have an N/A value. I have performed a differential protein analysis using limma which worked really well for each technical repeat however when combining all the biological repeats together the number of total of genes reduces dramatically as only genes that have no N/A values across all samples can be used.

Is there any way to work with what is there to expand the gene list out (i.e. still calculate p.value for those genes with only a few missing values)

Thanks again

[Ram]

Why did you delete the question, harrydolan.dc?

Answer 1

[Gordon Smyth]

You can use limma for mass spec data, but not voom. voom is designed specifically for count data and is neither needed for appropriate for mass spec data. If you use non-voom limma, then there is no need to remove NAs, so the problem you mention will go away. I would input log-expression values to limma then set trend=TRUE when you run eBayes.