fits latent class models for phenotypic measurements in pedigrees with or without familial dependence using an Expectation-Maximization (EM) algorithm