Unformatted text preview: as no
way of ‘remembering’ what a distant state
Sometimes, one can bend the rules of
HMMs without breaking the algorithms.
For instance, in genefinding, one wants to
emit a correlated triplet codon instead of
three independent residues; HMM algorithms can readily be extended to tripletemitting states. However, the basic HMM
toolkit can only be stretched so far. Beyond
HMMs, there are more powerful (though
less efficient) classes of probabilistic models
for sequence analysis.
1. Rabiner, L.R. A tutorial on hidden Markov models and
selected applications in speech recognition. Proc.
IEEE 77, 257–286 (1989).
2. Durbin, R., Eddy, S.R., Krogh, A. & Mitchison, G.J.
Biological Sequence Analysis: Probabilistic Models of
Proteins and Nucleic Acids (Cambridge University
Press, Cambridge UK, 1998). j Any model that has these properties is an HMM.
This means that one can make a new
HMM just by drawing a picture corresponding to the problem at hand, like
Figure 1. This graphical simplicity lets one
focus clearly on the biological definition of
a problem. Wondering how some other
mathematical technique really works?
Send suggestions for future primers to
[email protected] VOLUME 22 NUMBER 10 OCTOBER 2004 N ATURE BIOTECHNOLOGY...
View Full Document
- Spring '09
- DNA, Probability theory, Markov chain, Hidden Markov model, Markov models, state path