org.biojava.bio.dp
Interface EmissionState
- All Known Implementing Classes:
- SimpleEmissionState
- public interface EmissionState
- extends State, Trainable
A state in a markov process that has an emission spectrum.
These states have an associated Distribution. Within an HMM, these are the
states that actualy make your observed sequence. They also must supply
training behaviour to set the emission spectrum up.
- Author:
- Matthew Pocock
getAdvance
public int[] getAdvance()
- Determine the number of symbols this state advances along
one or more symbol lists. In the simple case, this method
should almost always return {1} if it is a true `emmision'
state, or {0} if it is a dot state which only emits a gap
character. For pairwise HMMs, it will normally return {1, 1}
for match state, and {0, 1} or {1, 0} for a gap state. Under
some circumstances it may be valid to return values other
than 1 or 0, but you should consider the consequences for
HMM architecture very carefully.
Note that the int array returned by this method should
never be modified.
getDistribution
public Distribution getDistribution()
- Get the Distribution associated with this state.
If the state is to be added to an HMM, then the state's emission spectrum
must be compatible with the HMM - that is, their emission alphabets must
match.
- Returns:
- the current Distribution object used by this state
setDistribution
public void setDistribution(Distribution dis)
throws java.lang.UnsupportedOperationException
- Set the Distribution associated with this state.
- Parameters:
dis
- the new Distribution to use