/BEEMM

Model HMM pour la détection des isochores

Primary LanguageR

BEEMM

To do list

build_markov

Julie, Matilda, Léo, Laure

build_markov(sequence_table, order)`
  • return : une table kmer_proba

markow_likelihood

Miléna, Lise, Dorine, Manon

markow_likelihood(sequence, table_kmer_proba, log=TRUE)`

markov_likelihood_seq

Lisa, Océane, Justine

markov_likelihood_seq(sequence_table, table_kmer_proba)`

ou

markov_likelihood_seq(sequence_table, ...)`

markov_bayes

markov_bayes(M0, M1, sequence_testing, priorM0 = 0.5)
  • Applies markov_likelihood_seq on every M0 and M1
  • Computes p(M|Seq) using Bayes
  • Returns a tibble similar to markov_likelihood_seq with four extra columns
    • lprob_M0 and lprob_M1
    • lmodel_M0 and lmodel_M1

Julie, Laure, Léo

markov_specificity

markov_specificity(M0s, M1s, sequence_testing, priorM0 = 0.5)
  • Calls markov_bayes
  • Computes specificity for all model pairs
  • returns
  • Miléna, Lise*

markov_sensibility

function prototype:

markov_sensibility(M0s, M1s, sequence_testing,  priorM0 = 0.5)

Computation algorithm

  • Computes sensibility for all model pairs by calling markov_specificity 
    markov_specificity(M0s = M1s, M1s = M0s, sequence_testing, priorM0 = 1 - priorM0)

markov_validate

markov_validate(sequence_learning_M0, sequence_learning_M1,
                sequence_testing_M0, sequence_testing_M1,
                order_min, order_max,
                priorM0 = 0.5)
  • prepare the models lists
  • call markov_specificity & markov_sensibility
  • merges both results in a three columns table

Manon, Dorine

markov_validate_kfold

markov_validate_kfold(sequence_learning_M0, sequence_learning_M1,
                      order_min, order_max,
                      priorM0 = 0.5,
                      learning_fraction = 0.8, nrand = 10)
  • Repeat nrand times :
    • splits learning data set in learning and testing data sets
    • calls markov_validate
  • returns the concatenated results of the nrand estimates.

Lisa, Justine, Oceane

viterbi

Eric