(Some) large trees fail in ape 5.5

[erikrikarddaniel]

After updating from ape 5.4 to 5.5 the read.tree function fails on large trees (e.g. the attached) with the following message:

numbers of left and right parentheses in Newick string not equal

I have checked the tree, and parentheses are balanced.

Some time ago, I reduced a tree that failed and at some point read.tree managed to read it, which is why I think it has to do with size.

Downgrading ape to 5.4 solves the issue.

gtdbtk.ar122.classify.tree.gz

[emmanuelparadis]

Yes, your tree is correct, but ape has difficulties handling it because some quoted labels include a semicolon which read.tree identifies as the end of the tree. The way quoted labels are treated is not trivial: the code in ape 5.4 had problems with some other trees; see issue #1. There is a fix in ape 5.5-1 but it also fails with your tree.

I'm working on a solution that will work for all trees.

[erikrikarddaniel]

Thank you! I have personal experience with funny characters in names, so you have my full understanding! :-) I'll continue with 5.4 for the time being.

/D

[emmanuelparadis]

I found a solution that should please everybody. I pushed it here, so you can install version 5.5-1. If you prefer to test it without re-installing the whole package, you can do these three commands from R:

source("https://raw.githubusercontent.com/emmanuelparadis/ape/master/R/read.tree.R")
.treeBuild <- ape:::.treeBuild
.cladoBuild <- ape:::.cladoBuild

The last two lines are because these two functions are not exported by ape.

I'll check the code again later.

[erikrikarddaniel]

OK. I tried to install 5.5-1 with devtools::install_version() but it wasn't found. I suppose you're still working on it and haven't released it.

After executing your code, with 5.4 loaded, read.tree still works.

In any case, 5.4 works fine for me, so I'll wait until there's an official 5.5-1 released.

Thanks for looking at this so quick!

/D

[emmanuelparadis]

I made another small modification which should fix the last problems. Here's an example to illustrate the behaviour of read.tree; the file tr.tre is:

(a  b:1,'; : (
)':1)'[:;(")]':1;

This is read as:

> tr <- read.tree("tr.tre")
> tr

Phylogenetic tree with 2 tips and 1 internal nodes.

Tip labels:
  ab, '; : ()'
Node labels:
  '[:;(")]'

Rooted; includes branch lengths.

> tr$tip.label
[1] "ab"       "'; : ()'"

> tr$node.label
[1] "'[:;(\")]'"

To summarize:

• all line breaks are ignored;
• spaces (and TABs) in non-quoted labels are ignored;
• labels within (single) quotes can have all characters except single quotes.

For the moment, double quotes (or other fancy quotes) are not considered as quoting labels.