Biology Reference
In-Depth Information
We only have data for the leaves so that the branching process is implic-
itly assumed not to be observable and its reconstruction is our main goal.
This chapter is organized into three sections, progressing from the
general to the specific. In the first section, we give some overall features
used to classify phylogeny reconstruction methods and present some
popular methods. The second section treats a particular class, the opti-
mality-based methods. We present a general strategy to search for the
optimal tree(s) for a given criterion. One particular optimality-based
method is the least squares method. In the third section, we describe the
least squares tree building program developed in our research group. The
section also includes a time comparison with the corresponding imple-
mentation in PAUP*
2
and a detailed description of a new heuristic. We
conclude this chapter with an outlook on the assessment of the accuracy
of constructed trees.
2. Types of Tree Building Methods
Tree building methods can be classified according to several criteria. A
criterion can refer not only to the algorithmic or statistical method used
to infer a tree from given input data, but also to the input data itself.
Maximum likelihood (ML) methods, for example, get their name from
the statistical approach taken; whereas the term “distance methods”
refers to the type of input data. Certain criteria are related to each
other. For instance, ML and parsimony methods are both character-
based and optimize for a defined score, so we can expect that the two
methods use, to some extent, similar algorithmic approaches. In this
section, we first list prevalent criteria used to classify phylogeny meth-
ods. Subsequently, we present some widely used tree building methods
in light of these criteria.
•
Type of input data
. Sequence-based phylogeny methods normally
use a multiple sequence alignment (MSA) as input. The two-step
approach, MSA construction followed by phylogeny construction, is
a very common one. Nevertheless, it is well known that MSA and
phylogeny construction are closely related tasks; this can be seen in
