If you want to relate expressions among several groups, it's best to show dCt. If the comparisons at all done to the same reference, best show ddCt.
Never ever use simple barcharts (as used so terribly often in biomedical papers). If you have small sample sizes, you can show dCt values simply in 1D scatterplots. For larger sample sizes, dCt values can be shown in boxplots, or in dot-plots with error bars (indicating median and IQR or the mean and 95%CI). In case of ddCt values the only option are dot-plots with mean and CI (since there are no "individual measurements" and you can only provide the mean ddCt).
If you are forced to show 2dCt or 2ddCt, then I'd suggest to calculate all statistics (like medians, IQRs, means, CIs) for the dCt (or ddCt) values and potentiate these results to show them in a plot. For the means and CIs this gives you the "geometric means" with according CI (what is not symmetric around the mean).
If you are forced to use barcharts, you cannot show dCt because this quantity has no interpretable zero value (so a hight of a bar, even the direction [positive or negative] provides no interpretable information; so the purpose of the barchart [=showing bar areas] is completely off-topic). You can neither 2dCt or 2ddCt on a barchart, too, because here (1) not the zero but the on is the reference value, and even if you would place the x-axis at y=1, (2) the bar areas to not represent the importance or magnitudes of the effects (4-fold down is a tiny area, indicating that there is not much, 4-fold up is a huge bar, indicating that therer is a lot. But the biologically relevant information - what is obscured rather than transported - would be that both changes may be of equal importance).
Better avoid barcharts anyway and generally.
and the great books from Edward Tufte, in particular
"The Visual Display of Quantitative Information" and "Envisioning Information".
Souce: NovoPro 2018-02-22