Compare two or more phylogenetic trees

Computes a small set of standard metrics for comparing trees that come from different backends (or different runs of the same backend). Designed for the common case of "I retrieved a tree from rotl and another from fishtree — do they agree?"

Usage

pr_tree_compare(..., prune_to_common = TRUE)

Arguments

...: Two or more phylo objects, or two or more pr_tree_result objects (the tree slot is extracted), or multiPhylo objects (the first tree is used). Trees can be passed as positional arguments or as a named list.
prune_to_common: Logical. Restrict each tree to the shared tip set before computing topology metrics? Default TRUE — without this, RF distance is undefined when tip sets differ.

Value

A list with class pr_tree_compare and components:

n_trees: Number of input trees.
tip_sets: Named list of character vectors, one per tree.
shared_tips: Tips present in every input tree.
unique_to: Named list, one per tree, of tips present in that tree but not in every other tree.
n_shared: Length-1 integer.
pairwise_jaccard: Square matrix; (i, j) is the Jaccard index of tip_sets[[i]] vs tip_sets[[j]].
pairwise_rf: Square matrix of Robinson-Foulds distances between pairs of trees pruned to shared_tips. NA when the pair has < 4 shared tips.
pairwise_branch_cor: Square matrix of Pearson correlations between matching edge lengths in each pair, or NA when one or both trees have no branch lengths.

Details

RF distance is computed via ape::dist.topo() with the default method. Branch-length correlation matches edges by their tip-set bipartition: for each edge in tree A, the corresponding edge in tree B (if any) is the one that splits the same set of tips. The Pearson correlation is taken over the matched edge-length pairs; edges whose bipartition is absent in the other tree are dropped. This is a proper bipartition-matched correlation as introduced in Kuhner & Felsenstein (1994) for tree comparison.

References

Kuhner, M. K., & Felsenstein, J. (1994). A simulation comparison of phylogeny algorithms under equal and unequal evolutionary rates. Molecular Biology and Evolution 11(3): 459–468. doi:10.1093/oxfordjournals.molbev.a040126

Robinson, D. F., & Foulds, L. R. (1981). Comparison of phylogenetic trees. Mathematical Biosciences 53(1–2): 131–147. doi:10.1016/0025-5564(81)90043-2

Examples

# Two trees with identical tip sets
set.seed(1)
t1 <- ape::rtree(10)
t2 <- ape::rtree(10, tip.label = t1$tip.label)
cmp <- pr_tree_compare(t1, t2)
cmp$n_shared
#> [1] 10
cmp$pairwise_rf
#>       tree1 tree2
#> tree1     0    16
#> tree2    16     0

# Two trees with overlapping but not identical tips
t3 <- ape::rtree(8, tip.label = t1$tip.label[1:8])
cmp <- pr_tree_compare(t1, t3)
cmp$pairwise_jaccard
#>       tree1 tree2
#> tree1   1.0   0.8
#> tree2   0.8   1.0