Multiple Sequence Alignment with DIALIGN - Multiple Sequence Alignment Methods - page 183

searching for local pairwise sequence similarities, and by then
combining these “partial” similarities in a final multiple alignment.
The idea was that pairwise or MSA should be limited to those parts
of the sequences that share some statistically significant similarity,
but one should not try to align non-related parts of the sequences.
In this sense, DIALIGN is a sort of hybrid approach, between
local and global alignment methods. If sequence homology extends
over the entire length of the input sequences, it produces global
alignments similar to the output of strictly global methods such as
CLUSTAL , but if sequences share only one region of homology,
this region is aligned and the rest of the sequences is ignored.
A main advantage of DIALIGN is that the program itself decides
which parts of the sequences are to be aligned, so the user does not
need to decide if a local or a global methods is suitable for a
particular sequence set under study. One single program can be
applied to both global and local homologies. More importantly,
DIALIGN can deal with sequences that share regions of detectable
homology that are separated by unrelated regions. Such sequences
can neither be aligned by purely global MSA methods nor by purely
local motif finders. Nevertheless, although DIALIGN integrates
local and global MSA features in a single program, the main
strength of the program is local homology detection.
DIALIGN can be applied to nucleic acid and protein
sequences. Originally, the program was developed and tested with
a focus on protein sequences. However, when more and more
genomic DNA sequences became available around the turn of the
millennium, the hybrid strategy of DIALIGN combining local and
global alignment was found particularly useful to study genomic
sequences where conserved functional regions such as exons or
regulatory sites are separated by large non-related parts of the
genome.
With the increasing number of partially or fully sequenced
genomes, cross-species alignment of genomic sequences became a
valuable tool to discover functional elements in genome sequences,
see , e.g., [ 12 ]or[ 13 ] for a review. The idea is that functionally
conserved elements of the genome are more conserved than non-
functional parts, so local conservation in an alignment of syntenic
genomic sequences usually indicates biological functionality.
Obviously, neither traditional methods for global alignment
such as CLUSTAL W nor purely local methods as BLAST are
suitable to align genomic sequences. Although DIALIGN was
originally designed as a generic MSA program, not specialized for
genomic alignment, it was the first MSA method that could pro-
duce meaningful alignments of genomic sequences. In a number of
pioneering studies, DIALIGN was used to discover non-coding
functional elements in genome sequences that could later be con-
firmed in experimental studies, e.g., [ 14 , 15 ].