[BioC] sva + voom + limma. eBayes(trend=TRUE)

[Meritxell Oliva]

Hi Gordon,

Thanks for your quick answer.

I used trend=TRUE 'cause you posted in Bioconductor that it's a good strategy if you are not filtering for genes with low counts,  as this strategy  is devoted  to model a good mean-variance trend for genes on the low expression range. This is the message I got, but correct me if I am wrong, please. Post: https://stat.ethz.ch/pipermail/bioconductor/2013-May/052816.html

If, as you say, voom already handles this issue, excellent, there is no need to set trend=TRUE in eBayes function. 
However, if I have already performed the analysis with trend=TRUE in eBayes function, do you think that this bears heavily on the DE analysis results? If so, the bias may go in which direction: increasing the number of false positives or false negatives?  I'm asking that for practical reasons, so as to decide to repeat the analysis again or use the results I have already produced.

Thanks a lot 

Meri

Meritxell Oliva
PhD student 
IBB (Biotechnology and Biomedicine Institute)
Comparative and Functional Genomics group
Campus Universitari - 08193 Bellaterra Cerdanyola del Vallès - Barcelona