Ggtree fails to display boxplots with high values

[seth-1]

Hi! Thanks for creating this great package!

I tried to combine a boxplot and a tree using facet_plot. Unfortunately, when displaying boxplots with high values, the plot seems to adjust ylim automatically so the plot 'disappears'.

tr <- rtree(30)
p <- ggtree(tr)
d4 = data.frame(id=rep(tr$tip.label, each=20), val=as.vector(sapply(1:30, function(i) rnorm(20, mean=i))) )
# Works fine
facet_plot(p, panel="Boxplot", data=d4, geom_boxploth, mapping = aes(x=val, group = label))

d5 <- d4
d5$val <- d5$val*100
# Disappears to the bottom of the device
facet_plot(p, panel="Boxplot", data=d5, geom_boxploth, mapping = aes(x=val, group = label)) 

Correct plot

Compressed plot

[GuangchuangYu]

> p <- ggtree:::add_panel(p, 'Boxplot')
> df = p %+>% d5
> head(df)
  label branch.length y isTip  .panel      val
1    t1     0.1369876 4  TRUE Boxplot 2722.792
2    t1     0.1369876 4  TRUE Boxplot 2606.299
3    t1     0.1369876 4  TRUE Boxplot 2707.818
4    t1     0.1369876 4  TRUE Boxplot 2769.059
5    t1     0.1369876 4  TRUE Boxplot 2623.170
6    t1     0.1369876 4  TRUE Boxplot 2845.702
> ggplot(df) + geom_boxploth(aes(x=val, y=label, group=label)) -> p1
> ggplot(df) + geom_boxploth(aes(x=val, y=y, group=label)) -> p2
> multiplot(p1, p2, ncol=2)

This seems to be an issue of ggstance. Maybe @lionel- can help.

[lionel-]

It seems to work fine with latest ggstance/ggplot2.

[wcornwell]

I am also having this issue with these versions:

ggtree_1.7.4
ggplot2_2.2.0.9000
ggstance_0.3

not sure what's going on but I'm getting the same behaviour as @seth-1 describes.

[Andreas-Bio]

I get the same error. A temporary workaround that works for me is to divide all data in the right panel by 10000.

My guess is that the y coordinate is not transferred correctly. Because if the right figure data goes larger as the number of tips it goes crazy. If it is smaller it is fine.

ggstance_0.3
ggtree_1.6.5 
ggplot2_2.2.0

data[,2] <- data[,2]/10000

[lionel-]

could you post your R version please? ggstance fails CRAN's checks with old versions of R, this may be related.

[Andreas-Bio]

sessionInfo()
R version 3.3.2 (2016-10-31)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:
[1] LC_COLLATE=German_Germany.1252 LC_CTYPE=German_Germany.1252 LC_MONETARY=German_Germany.1252
[4] LC_NUMERIC=C LC_TIME=German_Germany.1252

attached base packages:
[1] stats4 parallel stats graphics grDevices utils datasets methods base

other attached packages:
[1] ggstance_0.3 ggtree_1.6.5 ggplot2_2.2.0 Biostrings_2.42.1 XVector_0.14.0
[6] IRanges_2.8.1 S4Vectors_0.12.1 BiocGenerics_0.20.0 rentrez_1.0.4 taxize_0.8.0
[11] magrittr_1.5 ips_0.0-7 XML_3.98-1.5 colorspace_1.3-2 ape_4.0

[lionel-]

Hi all. Could you confirm that with the latest ggstance version (the one on github as it has not yet been sent to CRAN), the issue is fixed?

[wcornwell]

The scaling issue looks like resolved, at least for me. Thanks!

By the way, is there any way to get an axis on the horizontal boxplot atm?

[lionel-]

I'm not sure what you mean? But anything that is possible with vertical boxplots should be possible with horizontal ones, otherwise it's presumably a bug.

[GuangchuangYu]

@lionel- here is another example, https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/bioc-ggtree/RgI8M4DDBT8/7Kp-hyNqAQAJ.

Although y is correct, as shown in layer_data(p, 5)$y %>% range, without ylim the figure can't render properly.

[lionel-]

Can you open an issue with a small reprex at the ggstance repo please.

[GuangchuangYu]

good news from @lionel- that the issue has been fixed.