ggplot legend list is larger than page

Question

I have a plot in R which has a very large number of sample groups, and therefore the legend is larger than the page size and is cut off. I understand that this is not publication quality, but I need to know the colours to be able to make the legend in Illustrator.

Is there a way to make the page size much bigger or somehow change the legend format so that I can include all the keys? The reason for this is so that I can open the PDF in Illustrator and get the colours for each sample to create a new legend that will be for publication. I thought that maybe there is a clipping mask, and that the actual legend will be preserved, but when I opened in Illustrator, the legend was actually cut at the page ends1.

As was suggested in the comments below I gave nrow a try which helped break the legends up but now the entire page is just legends.

ggplot(purine.n, aes(x=variable, y=value, colour=metabolite_gene, shape=variable))
+geom_abline(slope=0)
+geom_point(size=4, position=position_dodge(width=0.08))
+scale_y_continuous(limit=c(-3.5,5.5), breaks=c(-3,-2,-1,0,1,2,3,4,5))
+scale_shape_manual(values=c(16,17,17), guide=F)
+theme_bw()
+theme(legend.key=element_blank(), legend.key.size=unit(1,"point"))
+guides(colour=guide_legend(nrow=16))

Disaccustom

Does using the nrow or ncol argument help? (imo (fwiw) having this many factors / colours is not going to produce a very informative plot / legend) – Badoglio

Do you need to have a different color for each category? Unless the categories are ordinal, this could be very hard for someone to read effectively. You may want to figure out a different way of differentiating categories. – Combine

Thanks, that led to this #25622420 but unfortunately it did not resolve the issue. I think what you pointed out is the right track tho! I've updated the question to reflect your suggestions. – Disaccustom

@user1362215 I do need the colours as I use the colours to seperate out subcategories within the sample. Actually, that will be the next thing I need to figure out is how to get category A and it's subcategories one shade of colour, B and it's subcategories another, and so on so forth. – Disaccustom

@Badoglio nrows worked but now I have a different problem. The entire page is just legends >.< – Disaccustom

@kennethPhough, ggplot is kind of bad when it comes to non-intuitive color schemes. You can use scale_color_manual, aided with some color-handling functions, to work with them. hsv2col and col2hsv might be useful in this case. – Combine

@user1362215 Ok, thanks for the advice! I'll definitely give that a try. – Disaccustom

@Badoglio Yes, thank you! The rnow attribute worked! – Disaccustom

Answer 1

As was suggested in the comments, nrow was the answer to my problem. I had to adjust the value to get the right number of rows to fit my legend. Below is the completed code that worked. There's more tweaking I need to do, like change page size to help make things look better, but that is out of the scope of this question.

ggplot(data.n, aes(x=variable, y=value, colour=metabolite_gene, shape=variable))
+geom_abline(slope=0)+geom_point(size=4, position=position_dodge(width=0.08))
+scale_y_continuous(limit=c(-3.5,5.5), breaks=c(-3,-2,-1,0,1,2,3,4,5))
+scale_shape_manual(values=c(16,17,17), guide=F)
+theme_bw()
+theme(legend.key=element_blank(), legend.key.size=unit(1,"point"))
+guides(colour=guide_legend(nrow=30))

Disaccustom

Does this also work when you have guides for color and shape? I like to have a line for each (so instead of your 30, I would only need 2 - but cannot figure out how to do this.) – Mighty