Thank you for cleaning this up, it's looking much better.

The filled points are positive in sign, and empty are negative. We are interested in the magnitude of the rho-statistics and the sign of the points are only secondary. So have the y-axis in symlog scale isn't particularly helpful.

What is the difference between the filled and empty symbols?

The plots from the most recent version of the branch is attached here. For the sake of clarity, the different Rho statistics are plotted to two different files, instead of two panels in one figure.

It would be better if the textual labels on the plot did not overlap so that they are easily legible. There is also plenty of whitespace on the plot, both within the axes and without, please could you move the legend so that it does not overlap the data points. It would also probably be better if the rerun information was at the bottom of the plot, otherwise it looks like that is relevant to the difference between panels. There is a function in fakesAnalysis that adds information to the outskirts of the plot that might be of use, or might not, but you could look at it.