Biology Reference
In-Depth Information
Chapter 10
Phylogeny-aware alignment with PRANK
Ari L ยจ ytynoja
Abstract
Evolutionary analyses require sequence alignments that correctly represent evolutionary homology.
Evolutionary and structural homology are not the same and sequence alignments generated with methods
designed for structural matching can be seriously misleading in comparative and phylogenetic analyses. The
phylogeny-aware alignment algorithm implemented in the program
PRANK
has been shown to produce good
alignments for evolutionary inferences. Unlike other alignment programs,
PRANK
makes use of phylogenetic
information to distinguish alignment gaps caused by insertions or deletions and, thereafter, handles the two
types of events differently. As a by-product of the correct handling of insertions and deletions,
PRANK
can
provide the inferred ancestral sequences as a part of the output and mark the alignment gaps differently
depending on their origin in insertion or deletion events. As the algorithm infers the evolutionary history of
the sequences,
PRANK
can be sensitive to errors in the guide phylogeny and violations on the underlying
assumptions about the origin and patterns of gaps. These issues are discussed in detail and practical advice for
the use of
PRANK
in evolutionary analysis is provided. The
PRANK
software and other methods discussed here
can be found from the program home page at
http://code.google.com/p/prank-msa/
.
Key words
Phylogeny-aware alignment, Evolutionary sequence analysis, Insertions and deletions,
Character homology
1
Introduction
Multiple sequence alignment has a central role in evolutionary
sequence analysis, in some studies so central that the alignment
and evolutionary inferences should be performed simultaneously or
at least in a tightly coupled manner. The connection between
alignment and phylogeny inference was noticed early [
1
], but the
evolutionary consequences of it are still largely ignored by main-
stream alignment methods. A probable explanation for this over-
sight is the historical focus on protein alignments and extensive use
of structural benchmarks in the development and comparison of
the analysis methods. The use of these benchmarks has produced
great alignments for structural studies of proteins but, as noticed by
many users of the resulting methods, the very same alignments may
be unsuitable for evolutionary analyses.
