Hello,

I am trying to reproduce the analysis done in Chapter 18 ("Profiles of Unrelated Nigerian Individuals") present in the Limma User's guide. I typed the commands described and I got the same genes differentially expressed between male and females. Here you go the first row of tapTable command:

                                 Symbol Chr logFC AveExpr t P.Value adj.P.Val B

ENSG00000229807 XIST X -9.857 3.8076 -32.5 1.21e-44 2.10e-40 66.5 

I would like to get the mean expression in male and female group, so I take a look at v$E matrix (where v is the voomed object and v$E contains the normalized values of the genes in my samples) and first of all I calculate the mean expression of XIST among the samples. It matches (3.8076) with the one reported by topTable. Then, I calculate the mean of expression in male and female group and the fold change as mean_male - mean_female, but I got a slightly different FC from the one reported by topTable:

My calculation FC: -9.815

topTable log2FC: -9.857

Could you help me to figure out this difference? Is Limma supposed to calculate the logFC as mean_group1 - mean_group2 because the values are logged? 

Best,

Luciano

http://www.bioconductor.org/packages/release/bioc/vignettes/limma/inst/doc/usersguide.pdf