This function provides an estimation of a mixture of gamma distributions. The code is inspired from the varmixt package.
mixgamma(VAR, dfreedom, var.init, pi.init, nmixt, stop.crit, display = TRUE,
niter.max = 50000, criterion = criterion)vector of estimated variance.
degrees of freedom of the estimated variance.
vector of initial variances for the mixture.
vector of initial prior probability for the mixture.
integer : number of components in the mixture model.
the stopping relative precision limit for stopping EM algorithm.
boolean : should the result of model fitting be displayed on the screen ?
integer : maximum number of iterations of the EM algorithm.
criterion for the convergence of the EM algorithm : either "likelihood" for criterion base on loglikelihood either "parameter".
a LIST with the following components :
value of the BIC criterion
the probability of each variance component
variances
value of the criterion base on Loglikelihood
number of variance components
the matrix of posterior probability that a gene belongs to each variance component. One row per gene, one column per variance component.
the variance attributed to each gene according to the MAP rule - Delmar et al. (2005) JRSS
the variance attributed to each gene taking into account the tau values - Delmar et al. (2005) Bioinformatics