Next: Computing the Likelihood of
Up: No Title
Previous: Neighbor Joining
Maximum Likelihood
Given a tree, we often wish to have a statistical measure of how well it
describes our data. As we have seen earlier in the course, we can use the
likelihood score to estimate our hypothesis, which is in this case a phylogenetic tree T.
For a set of species with observed values M, we would choose the tree that maximizes
P(M|T). In the following section, we shall assume that the tree topology is
known, and show how to find the optimal branch lengths. To this end, we
will first demonstrate how to calculate the likelihood of a tree efficiently. We
will not discuss the issue of searching among tree topologies, which suffers from
the same difficulties we mentioned in the previous sections, although is not proven
to be NP-complete.
Itshack Pe`er
1999-02-18