Hello mixOmics team,
When plotting the Diablo output in a circosPlot, I get the names of some of my variables covered by the block color and seems to depend on the number of variables and the rotation angle of the labels.
no problem iif only few variables are ploted
but yes when plotting more variables
any suggestion how to fix that, and yes indeed, in the plot pane I have no problem, only when saving it with pdf() function
Interesting that the issue only arising when saving via
pdf(). Have you explored saving it into different file types? Or by using a different set of functions?
If you’re saying that it plots fine within RStudio, then if worse comes to worst, you can plot it in RStudio, maximise the
Plots window, take a screenshot and crop it. I understand this isn’t the best solution - but this might be a product of the
pdf() function rather than
Please let me know how you go. I’ll see if I can recreate your issue as well
Hi Max, thanks for the reply!
Now I am not sure anymore if the problem was only when using
pdf() because now it is also happening in the
RStudio Plots window. I tried to play with
par() (because I recently changed some parameters to allow proper plotting of large datasets), but still get the var names hidden by the green blocks if the window is small or saving with
pdf(). But clearly, it depends on the total number of variables plotted with
circosPlot()…if you could help me on that, I would be very grateful! It may be more an issue from my end (config of ploting window or something like that…)
Seeing as you’re plotting a large number of features and “the window is small”, an overlap of these segments is unavoidable. Additionally, if you have a ton of features, then it’s safe to assume that you’re looking at the structures and relationships in the data so individual names are likely less important.
Have you tried using the RStudio export feature (via the “Plots” window)? Or another exporting function (there are many)? Or only
Another recommendation is if you are wanting to examine specific features by name, then reduce the number of features displayed. It makes for a more informative plot depending on what you aims with the figure are.