Midpoint rooting tree causes ggtree to fail, reporting multiple roots about ggtree

[GuangchuangYu]

This should be a bug.

After my testing, I found that it's not a bug of ggtree.

You can plot the tree successfully using ggtree(y, ladderize=FALSE).

The issue is due to ggtree call ape::ladderize() and the output tree from phangorn::midpoint() is not compatible with ape::ladderize().

> y

Phylogenetic tree with 10 tips and 9 internal nodes.

Tip labels:
    t8, t1, t4, t6, t10, t3, ...

Rooted; includes branch lengths.
> y$edge
      [,1] [,2]
 [1,]   15    3
 [2,]   15    4
 [3,]   14    2
 [4,]   14   15
 [5,]   13   14
 [6,]   13    5
 [7,]   12   13
 [8,]   12    6
 [9,]   18    1
[10,]   18   12
[11,]   19    8
[12,]   19    9
[13,]   17   10
[14,]   17   19
[15,]   16    7
[16,]   16   17
[17,]   11   16
[18,]   11   18
> ladderize(y)

Phylogenetic tree with 10 tips and 9 internal nodes.

Tip labels:
    t8, t1, t4, t6, t10, t3, ...

Rooted; includes branch lengths.
> ladderize(y)$edge
      [,1] [,2]
 [1,]   11   18
 [2,]   18   12
 [3,]   16   17
 [4,]   17   19
 [5,]   19    8
 [6,]   19    9
 [7,]   15    4
 [8,]   17   10
 [9,]   18    1

The output from ape::ladderize() is not correct. We need to report this issue to the author of ape and phangorn.

[davidhwyllie]

Thank you very much for your reply.
Do you wish me to report this issue, or will you do so?
With all best wishes
David

From: Guangchuang Yu [mailto:[email protected]]
Sent: 04 October 2015 15:12
To: GuangchuangYu/ggtree [email protected]
Cc: David Wyllie [email protected]
Subject: Re: [ggtree] Midpoint rooting tree causes ggtree to fail, reporting multiple roots (#15)

This should be a bug.

After my testing, it's not a bug of ggtree.

You can plot the tree successfully using ggtree(y, ladderize=FALSE).

The issue is due to ggtree call ape::ladderize() and the output tree from phangorn::midpoint() is not compatible with ape::ladderize().

y

Phylogenetic tree with 10 tips and 9 internal nodes.

Tip labels:

t8, t1, t4, t6, t10, t3, ...

Rooted; includes branch lengths.

y$edge

  [,1] [,2]

[1,] 15 3

[2,] 15 4

[3,] 14 2

[4,] 14 15

[5,] 13 14

[6,] 13 5

[7,] 12 13

[8,] 12 6

[9,] 18 1

[10,] 18 12

[11,] 19 8

[12,] 19 9

[13,] 17 10

[14,] 17 19

[15,] 16 7

[16,] 16 17

[17,] 11 16

[18,] 11 18

ladderize(y)

Phylogenetic tree with 10 tips and 9 internal nodes.

Tip labels:

t8, t1, t4, t6, t10, t3, ...

Rooted; includes branch lengths.

ladderize(y)$edge

  [,1] [,2]

[1,] 11 18

[2,] 18 12

[3,] 16 17

[4,] 17 19

[5,] 19 8

[6,] 19 9

[7,] 15 4

[8,] 17 10

[9,] 18 1

The output from ape::ladderize() is not correct. We need to report this issue to the author of ape and phangorn.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/15#issuecomment-145353036.

[KlausVigo]

Dear Guangchuang & Emmanuel,

it seems ladderize assumes that trees are in "cladewise" order, but midpoint returns a tree in postorder.

x<-rtree(10)
x2 <- reorder(x, "postorder")
plot(ladderize(x2))  # error message

y<-phangorn::midpoint(x)
y <- reorder(y)
plot(ladderize(y)) # works!!!

Just add

tree <- reorder(tree)

before running ladderize.

Cheers,
Klaus

[GuangchuangYu]

Dear Klaus,

Thank you for your reply.

It's very clear. We just need to reorder() the tree before using ggtree().

Thanks,
Guangchuang

[GuangchuangYu]

@davidhwyllie FYI, In r-sig-phylo mailing list, Emmanuel mentioned that he will include phy <- reorder(phy) in ladderize(), so we don't need to call reorder explicitly in future release of ape.