Compute most probable path with extended Viterbi algorithm.

Share:

Description

Viterbi algorithm for Hidden Markov Model with duration

Usage

1
duration_viterbi(aa_sample, pipar, tpmpar, od, params)

Arguments

aa_sample

character vector representing single aminoacid sequence.

pipar

probabilities of initial state in Markov Model.

tpmpar

matrix of transition probabilities between states.

od

matrix of response probabilities. Eg. od[1,2] is a probability of signal 2 in state 1.

params

matrix of probability distribution for duration. Eg. params[10,2] is probability of duration of time 10 in state 2.

Value

A list of length four:

  • path a vector of most probable path

  • viterbi values of probability in all intermediate points,

  • psi matrix that gives for every signal and state the previous state in viterbi path,

  • duration matrix that gives for every signal and state gives the duration in that state on viterbi path.

Note

All computations are on logarithms of probabilities.