I have recently checked with the MAST developers what log basis they use to report logfoldchanges in MAST because I could not find that information anywhere. (RGLab/MAST#112)
Turns out that given we provide (natural)log-normalised as an input to mast, the reported logfoldchanges are also on a natural log basis. I.e. they need to be devided by ln(2) to become log2FC.
I guess this will also be important to how you use those values in the notebook as well.