log2FoldChange in DESeq2

[Golsheed]

Hello,
I have some trouble understanding how the log2FoldChange is calculated in DESeq2? Assuming that the full model is ~ AF + Condition, I'm interested in the log2FoldChange of the raw counts with respect to the covariate Condition (which is a factor with two levels).
I was under the impression that its calculated as the log2 of the ratio of mean of the normalized counts in the two levels. However, this is not what I get from DESeq2. Could someone please give me a hint on how these values are actually calculated?
Or is there a way I can get an estimate (or values close the log2FoldChange obtained from DESeq2) without using DESeq2?
Thanks,
Golsheed

[fanli]

Are you looking at the moderated fold changes or the unshrunken MLEs? See section 1.4.1:

http://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.pdf

You may want to do something like this to get the actual log2FoldChange values:

resMLE <- results(dds, addMLE=T)

[Michael Love]

In a previous thread, I answered one of your questions with " log2FoldChange with betaPrior=FALSE .... is the log2 ratio of the mean of normalized counts"

The default is to use a prior on the log2 fold change and give an estimate using the posterior distribution of log2 fold change.

See the paper for full details:

http://genomebiology.com/2014/15/12/550

...or check the help files: ?nbinomWaldTest.

If you use betaPrior=FALSE, then no beta prior is used. If you use addMLE=TRUE as fanli suggests, then a beta prior is used, but a column is added to the results table which gives the unshrunken estimate of log2 fold change which is mean(normalized counts)/mean(normalized counts). See ?results for full details on this.