Biology Reference
In-Depth Information
Chapter 12
Multiple Sequence Alignment with DIALIGN
Burkhard Morgenstern
Abstract
DIALIGN is a software tool for multiple sequence alignment by combining global and local alignment
features. It composes multiple alignments from local pairwise sequence similarities. This approach is
particularly useful to discover conserved functional regions in sequences that share only local homologies
but are otherwise unrelated. An anchoring option allows to use external information and expert knowledge
in addition to primary-sequence similarity alone. The latest version of DIALIGN optionally uses matches to
the
PFAM
database to detect weak homologies. Various versions of the program are available through
G¨ttingen Bioinformatics Compute Server (GOBICS)
at
http://www.gobics.de/department/software
.
Key words
Motif discovery, Local alignment, Anchored alignment, Protein domain
1
Introduction
When the first software tools for multiple sequence alignment
(MSA) were developed, alignment methods were generally classi-
fied as either
global
or
local
. Global alignment methods align
sequences over their entire length, regardless of the real extension
of sequence homology. Local methods return only one or several
regions of sequence similarity, ignoring the remainder of the
sequences. The classical
progressive
alignment methods proposed
in the late 1980s were global methods [
1
-
3
]. Until today, many of
the most advanced MSA approaches rely on this concept, e.g.,
MAFFT
[
4
],
MUSCLE
[
5
], T-COFFEE [
6
], or
CLUSTAL
Omega
[
7
]. In addition to these purely global approaches, algo-
rithms for
local
multiple alignment have been proposed since the
early days of MSA. Pioneering local MSA methods are
MEME
[
8
],
PIMA
[
9
], or
TEIRESIAS
[
10
]. These tools are limited to extract
conserved motifs from a set of sequences, even if detectable homol-
ogy is not restricted to one conserved motif.
The program
DIALIGN
was first released in 1996 [
11
], the
main idea behind this program was to combine local and global
alignment features. In
DIALIGN
, this is implemented by first
