Construct a consensus tree from a collection of trees

A consensus tree can be constructed from a collection of cogent3 tree objects with consensus_tree. The method defaults to the majority-rule consensus tree - though different minimum levels of support can be used to achieve the strict consensus tree, and extended majority-rule consensus tree. By default, the support values of the input trees are carried over onto the consensus tree (see Bootstrap values below).

Usage

Majority-rule consensus tree

From a collection of trees, the majority-rule consensus tree can be constructed.

from cogent3 import make_tree
from piqtree import consensus_tree

tree1 = make_tree("(a,(b,(c,(d,e))));")
tree2 = make_tree("(a,((b,c),(d,e)));")
tree3 = make_tree("(a,((b,d),(c,e)));")

tree = consensus_tree([tree1, tree2, tree3]) # (a,(b,c,(d,e)));

Strict consensus tree

The min_support parameter can be set to 1.0 to construct a strict consensus tree (only the clades that appear in all of the trees). The parameter represents the proportion of trees a clade must appear in to be in the resulting tree.

from cogent3 import make_tree
from piqtree import consensus_tree

tree1 = make_tree("(a,(b,(c,(d,e))));")
tree2 = make_tree("(a,((b,c),(d,e)));")
tree3 = make_tree("(a,((b,d),(c,e)));")

tree = consensus_tree([tree1, tree2, tree3], min_support=1.0) # (a,(b,c,d,e));

Extended majority-rule consensus tree

When the min_support parameter is equal to 0, the extended majority-rule consensus tree is computed.

from cogent3 import make_tree
from piqtree import consensus_tree

tree1 = make_tree("(a,(b,(c,(d,e))));")
tree2 = make_tree("(a,((b,c),(d,e)));")
tree3 = make_tree("(a,((b,c),(d,e)));")
tree4 = make_tree("(a,(c,(b,(d,e))));")
tree5 = make_tree("(c,(b,(a,(d,e))));")

tree = consensus_tree([tree1, tree2, tree3, tree4, tree5], min_support=0) # (a,((b,c),(d,e)));

Bootstrap values

When the input trees carry their own branch support - for example, the ultrafast bootstrap support produced by build_tree - consensus_tree carries these values over onto the matching clades of the consensus tree, averaged across the input trees that contain each clade. The way the values are combined is set by support_aggregate ("mean" by default, or "max" / "min"), and each node's value can be accessed from node.support.

from cogent3 import make_tree
from piqtree import consensus_tree

# the numbers on the internal nodes are each tree's own bootstrap support
tree1 = make_tree("(a,((b,c)96,(d,e)90));")
tree2 = make_tree("(a,((b,c)88,(d,e)70));")

tree = consensus_tree([tree1, tree2])  # support_aggregate="mean" by default
# (b,c) -> mean(96, 88) -> 92.0
# (d,e) -> mean(90, 70) -> 80.0