Biomedical Engineering Reference
In-Depth Information
and the relative position of each sequence is allowed to mutate and mate with other patterns.
Offspring of these original patterns that maximize the alignment score are in turn allowed to mate
and mutate, creating other patterns. In this way,
an
optimal—but not
the
optimal—multiple
alignment solution is obtained.
Multiple alignment methods based on HMMs have been incorporated into a variety of tools. As
introduced in
Chapter 7
, "Data Mining," a HMM is a statistical model for an ordered sequence of
symbols, acting as a stochastic state machine that generates a symbol each time a transition is made
from one state to the next. A limitation of a HMM approach is that the model must be trained before
it can be used. As such, HMMs tend to be problem-specific, albeit powerful.
Other Strategies
There are dozens of approaches to multiple sequence alignment, some relegated to specific
laboratories, and others vying for use as a standard in the bioinformatics arena. Many of these
methods are highly specialized at solving specific types of multiple sequence alignment problems. For
example, the eMOTIF Method is optimized for identifying motifs in protein sequences. Profile analysis
is used for localized alignments in multiple sequence analysis. BLOCK analysis is used for working
with conserved regions (blocks) in a multiple sequence alignment. Expectation Maximization (EM) is
used to perform local multiple sequence alignment (as in Multiple EM for Motif Elicitation or MEME).
These and other approaches are constantly evolving, thanks to feedback and support from the
worldwide bioinformatics user community.
