This preview shows page 1. Sign up to view the full content.
Unformatted text preview: s the six highest-scoring paths (those 1316 with G at the 5′SS). The best one has a log
probability of –41.22, which infers that the
most likely 5′SS position is at the fifth G.
For most problems, there are so many
possible state sequences that we could not
afford to enumerate them. The efficient
Viterbi algorithm is guaranteed to find the
most probable state path given a sequence
and an HMM. The Viterbi algorithm is a
dynamic programming algorithm quite
similar to those used for standard sequence
Beyond best scoring alignments
Figure 1 shows that one alternative state
path differs only slightly in score from putting the 5′SS at the fifth G (log probabilities
of –41.71 versus –41.22). How confident are
we that the fifth G is the right choice?
This is an example of an advantage of
probabilistic modeling: we can calculate our
confidence directly. The probability that
residue i was emitted by state k is the sum
of the probabilities of all the state paths
that use state k to generate residue i (that is,...
View Full Document
- Spring '09