Hi David,
It's a follow-up on your message.
David Williams wrote on 22/06/2011 07:31:
Dear R users and phylogeneticists,
I've been manipulating some phylo objects mostly with ape functions but
also with some phangorn functions. I'm a bit confused with some effects
of rerooting and the output of various functions. Any help would be
greatly appreciated!
R code:
## Rerooting seems to fail sometimes:
library(ape)
tree <-
read.tree(text="((((G:1,H:1)g:1,(((A:1,B:1)c:1,(C:1,D:1)d:1)b:1,(E:1,F:1)e:1)a:1)f:1,K:1)h:1,I:1,J:1)i;")
clan <- list(c("G","H"),c("A","B","C","D","E","F"))[[2]]
layout(matrix(c(1,2), 1, 2), width = c(1, 1))
plot(tree,'u');nodelabels()
treeR <- root(tree, outgroup = c("I","J","K"), resolve.root = F)
plot(treeR,'u');nodelabels()
treeR$edge[1]
reorder(treeR)$edge[1]
## the root node of treeR is unchanged and within the requested
## outgroup :-(
This was a bug which is now fixed:
https://svn.mpl.ird.fr/ape/dev/ape/R/root.R
plot(tree,'u');nodelabels()
treeR <- root(tree, outgroup = c("G","H"), resolve.root = F) #
put the root elsewhere temporarily
treeR <- root(treeR, outgroup = c("I","J","K"), resolve.root = F)
plot(treeR,'u');nodelabels()
treeR$edge[1]
reorder(treeR)$edge[1]
## the root node of treeR is now at the base of the outgroup as
## expected :-) but an additional rerooting was necessary.
That was indeed a work-around.
## all.equal.phylo() and therefore unique.phylo() are sensitive to edge
## order in the edge matrix in unrooted trees:
clado1 <-
read.tree(text="((:1,(:1,:1):1):1,:1,((:1,:1):1,((:1,:1):1,:1):1):1);")
clado2 <-
read.tree(text="((:1,(:1,:1):1):1,(:1,:1):1,(:1,((:1,:1):1,:1):1):1);")
plot(clado1,'u');tiplabels();nodelabels()
plot(clado2,'u');tiplabels();nodelabels()
## I would say these are topologically identical
is.rooted(clado1) # FALSE
is.rooted(clado2) # FALSE
## confirm these phylo objects are not rooted
all.equal.phylo(clado1,clado2,use.tip.label=F,use.edge.length=F) # FALSE
The difficulty is that these trees have empty strings as tip labels, so
the algorithm used by all.equal.phylo does not work in this situation.
However, this affects only the unrooted trees as you noted. I'm going to
add a warning in the help page to mention this fact.
Thanks for the report (and your patience).
Cheers,
Emmanuel
## all.equal.phylo thinks they are different . . . .
length(unique(c(clado1,clado2))) # 2
## things are as expected if a root is added in equivalent places:
library(phangorn)
plot(midpoint(clado1),'u');tiplabels();nodelabels()
plot(midpoint(clado2),'u');tiplabels();nodelabels()
## I would say these are topologically identical
## all.equal.phylo agrees this time
all.equal.phylo(unroot(midpoint(clado1)),unroot(midpoint(clado2)),use.tip.label=F,use.edge.length=F)
# TRUE
length(unique(c(midpoint(clado1),midpoint(clado2)))) #1
## if the phylo objects are topologically identical, I would expect
## a Robinson-Foulds distance of 0
RF.dist(clado1,clado2) # 4
RF.dist(reorder(clado1),reorder(clado2)) # 4
RF.dist(reorder(midpoint(clado1)),reorder(midpoint(clado2))) # 6
RF.dist(midpoint(clado1),midpoint(clado2)) # 6
Thanks!
--
Emmanuel Paradis
IRD, Jakarta, Indonesia
http://ape.mpl.ird.fr/
_______________________________________________
R-sig-phylo mailing list
R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
