ABSTRACT
Multiple sequence alignment is more complex and requires somewhat different tools than pairwise sequence alignment. Selection and implementation of costs to produce an alignment are partially arbitrary. Most alignment programs allow for insertion of user-defined alignment parameters. Alignments should use multiple optimality criteria to explore the alignment space. Choosing an alignment program is an important first step. While we have described a progressive-only algorithm in this chapter (CLUSTAL), we suggest that researchers explore other algorithms and approaches to alignment. The recent development of iterative approaches has rendered the progressive alignments nearly obsolete. Changing alignments can change the phylogenetic hypothesis obtained, and methods are needed to accommodate this phenomenon. Cull and elision are techniques that can be used to deal with variation of alignments. Alignment by eye, while appealing to our visual nature as humans, is not an acceptable approach unless every step of the ocular alignment is recorded and reported. Alignments by eye can rarely be replicated from one researcher to the next, and repeatability is a hallmark of good science.
Other methods that do not involve multiple sequence alignment, such as optimization alignment and fixed-state optimization, may be used to generate phylogenies
