Troubleshooting

Every entry here is a failure someone has actually hit with this package or with the packages it sits on. Nothing on this page is hypothetical.

could not find function "is.waive" when a tree is drawn

Symptom: plot_tb_tree() or plot_tb_cladogram() returns without complaint, and then printing or saving the result fails.

Error in `geom_segment2()`:
! Problem while converting geom to grob.
i Error occurred in the 4th layer.
Caused by error in `is.waive()`:
! could not find function "is.waive"

Cause: an old ggtree meeting ggplot2 4.x. ggplot2 4.0 renamed the internal helper is.waive() to is_waiver(), and ggtree's aligned tip labels still call the old name. Both tree helpers use geom_tiplab(align = TRUE), so both hit it.

This is not a mycolorsTB bug and mycolorsTB is not involved. It reproduces with plain ggtree, in a session where this package was never loaded:

library(ggtree)

tr <- ape::read.tree(text = "(A,(B,C));")

print(ggtree(tr) + geom_tiplab(align = TRUE))   # fails
print(ggtree(tr) + geom_tiplab(align = FALSE))  # renders

Fix: the honest answer is that BiocManager::install("ggtree") on its own is probably not it, and it is worth knowing why before you run it. Bioconductor releases in lockstep, and BiocManager picks the release your R version is entitled to. On R 4.4 that is Bioconductor 3.20, whose newest ggtree is 3.14.0, which is the version that calls is.waive(). So the obvious command reports that there is nothing to do:

BiocManager::install("ggtree")
#> Bioconductor version 3.20 (BiocManager 1.30.25), R 4.4.2 (2024-10-31)
#> ... warning: package(s) not installed when version(s) same as or greater
#> than current; use `force = TRUE` to re-install: 'ggtree'

Check where you stand, all four together, because the pairing is what matters and no one of them tells you anything alone:

R.version.string
BiocManager::version()
packageVersion("ggtree")
packageVersion("ggplot2")

The upgrade that reaches a fixed ggtree is R plus Bioconductor, not ggtree on its own: move to an R version whose Bioconductor release carries a ggtree that has caught up with the rename, then reinstall the Bioconductor stack against it. BiocManager::install(version = "...") refuses a release your R cannot have, and the refusal is the check that tells you which way round to do it:

BiocManager::install(version = "3.22")
#> Error: Bioconductor version '3.22' requires R version '4.5'; use
#>   `version = '3.20'` with R version 4.4; see https://bioconductor.org/install

If you cannot move R, the remaining way out is the other half of the pairing: hold ggplot2 below 4.0, where is.waive() still exists. That pins you behind the rest of the ecosystem, so it is a way to keep working rather than a fix. The align = FALSE escape in the reproducer above is not available through this package: plot_tb_tree() and plot_tb_cladogram() each take newick_text and nothing else, and align = TRUE is written into the call they build.

My bars came out grey

Symptom: a bar chart or a scatter plot where some categories carry palette colours and the rest are grey, with no error, no warning, and no legend entry for the grey ones.

Cause: mycolors is a named vector, and the four *_mycolors() scales match by name. A value that is not one of A1 to A4 or L1 to L10 is treated as missing and filled with na.value, whose default is "grey50". Nothing is reported, because from ggplot2's point of view nothing went wrong.

Confirm it in one line:

counts <- data.frame(lineage = c("L2", "L4", "Beijing", "unknown"), n = c(112, 208, 40, 6))

setdiff(unique(counts$lineage), names(mycolors))
#> [1] "Beijing" "unknown"

character(0) means every value has a colour. Anything else is exactly the set of values that came out grey. The usual culprits are sublineage codes (L4.9), spoligotype family names (Beijing, LAM), lowercase (l4), leading spaces, and placeholders such as unknown, NA and -.

Fixes, in order of preference:

1. Recode the column to the fourteen names the palette knows.
2. If some rows genuinely have no lineage, keep them but make the fallback deliberate: scale_fill_mycolors(na.value = "grey20") and say in the caption what it means. na.value became settable in 0.1.2.
3. If the variable is not lineages at all, use scale_fill_classicTB(), which colours by position and never leaves a level uncoloured, or build the mapping yourself with scale_fill_manual().

The same trap applies to trees, where the tip labels are the values being matched. See Trees.

Insufficient values in manual scale

Symptom:

Error in `palette()`:
! Insufficient values in manual scale. 18 needed but only 14 provided.

Cause: the classicTB scales carry 14 colours and mycolors carries 14 names. A manual scale in ggplot2 needs one value per level and refuses to recycle. The error arrives at draw time, so the plot object is built successfully and fails when you print or save it.

Fix: generate as many colours as you have levels and pass them to a plain manual scale.

many <- data.frame(region = sprintf("R%02d", 1:18), n = seq(30, 200, length.out = 18))

k <- nlevels(factor(many$region))

ggplot(many, aes(x = region, y = n, fill = region)) +
  geom_col() +
  scale_fill_manual(values = tb_palette(k, "classicTB"), guide = "none")

tb_palette() will warn that it interpolated. Read the warning rather than suppressing it: interpolated colours are not palette colours, and more than fourteen categories is more than a reader can tell apart on any palette. Merging the small categories into an "other" bucket usually makes a better figure than eighteen shades.

install.packages() did not get me ggtree

Symptom: loading mycolorsTB fails with

there is no package called 'ggtree'

and install.packages("ggtree") does not fix it, warning instead that the package is not available.

Cause: ggtree is a Bioconductor package, not a CRAN one, and install.packages() only looks at the repositories configured for it, which by default are CRAN's. mycolorsTB declares ggtree in Imports, so the dependency is real and the load fails without it.

Fix:

if (!requireNamespace("BiocManager", quietly = TRUE)) {
  install.packages("BiocManager")
}
BiocManager::install("ggtree")

Then install mycolorsTB as usual. Only plot_tb_tree() and plot_tb_cladogram() need ggtree; the palettes and the four ggplot2 scales do not, but R loads the whole namespace, so the package will not attach until ggtree is present.

If BiocManager::install() reports that your Bioconductor release does not match your R version, upgrade R first. Bioconductor pins a release to an R version and will not install a package built for another one.

The same script gives different colours than it used to

Three behaviours changed in 0.1.2. One changes what a successful call returns; the other two turn something that used to succeed quietly into an error. Either way, a script that ran on 0.1.1 behaves differently on 0.1.2. Check with packageVersion("mycolorsTB").

tb_palette() no longer interpolates below the palette size

On 0.1.1 every call went through colorRampPalette(), whatever n was. On 0.1.2 the palette colours are returned unchanged while n does not exceed the palette size, and interpolation happens only above it.

tb_palette(5, "mycolors")
#> [1] "#d1ae00" "#8ef5c8" "#73c2ff" "#ff9cdb" "#ff3091"

Those are now A1, A2, A3, A4 and L1. On 0.1.1 they were five points sampled along a ramp through all fourteen, matching no lineage, while the function's own warning said interpolation only happened when you asked for more colours than the palette held.

If a figure has to keep the old colours, the old behaviour is one call: grDevices::colorRampPalette(mycolors)(5). In almost every other case the new colours are the ones you meant.

Interpolated output is uppercase and palette colours are lowercase, which is a quick way to tell which you are looking at.

A number in palette_name is now an error

tb_palette(3, 2)
#> Error: `palette_name` must be a single palette name, one of: mycolors, classicTB, pathogenomics.

On 0.1.1 that reached switch(), which selects by position when it is handed a number, and returned colours from the second palette without saying so. The argument order is tb_palette(n, palette_name), so a lone palette name is also caught now:

tb_palette("classicTB")
#> Error: `n` must be a single non-negative whole number.

Negative, fractional, missing and infinite values of n are rejected the same way, where before they either rounded silently or failed with an internal message from colorRampPalette().

Malformed Newick is now an error

plot_tb_tree("(L1,L2")
#> Error: `newick_text` could not be parsed as a single tree in Newick format.

On 0.1.1 ape::read.tree() returned NULL for this and the function carried on, producing a plot with an empty axis range: a blank panel and no explanation. Any pipeline that was quietly generating blank tree panels will now stop at the tree that caused it.

There is no scale_fill_pathogenomics()

Only mycolors and classicTB have scale functions. The pathogenomics palette is exported as a plain vector and reached directly:

df <- data.frame(group = c("cases", "controls", "unresolved"), y = c(64, 51, 12))

ggplot(df, aes(x = group, y = y, fill = group)) +
  geom_col() +
  scale_fill_manual(values = pathogenomics)

or through tb_palette(n, "pathogenomics") when you want a specific number of colours. Bear in mind that it is a theme palette running from near-white to near-black, so it is not eight mutually distinguishable categorical colours.

Nothing here matches

Open an issue at PathoGenOmics-Lab/mycolorsTB with the output of sessionInfo() and the smallest script that reproduces the problem. For anything that draws a tree, the versions of ggtree and ggplot2 are usually the answer, so include them.