MODEL OF MIXTURE OF DISTRIBUTIONS FOR THE EFFECTS OF GENETIC CHANGES IN THE TUMOR PROGRESSION OF BREAST CANCER
Bimodality, Gene expression, Breast, Prognosis, Clinical Profiles.
Several models have been developed with the purpose of identifying more homogeneous subgroups of patients presenting prognostic distinctions, aiming to improve the therapeutic approach and its efficiency for the respective subgroups selected. Genes that present differentiated expression values in subgroups of patients construct what we call bimodal expressions of genes. The set of these continuous expression values (low and high), for the same gene, has a definition of bimodal distribution vague, the term typically refers to the mixture of Gaussian distributions. The main motivation for the use of a mixture of distributions in genetics is the possibility of modeling or accommodating the heterogeneity of expression values to minimize the impact of underlying genetic factors. Bessarabova et al. (2010) emphasizes that the distribution with two distinct peaks means that the transcriptional regulation of some genes is conditioned to the type of breast cancer, so the subcategorization of cancer patients is extremely difficult due to the high heterogeneity of expression profiles. We propose a methodology that identifies these genes with bimodal density, with the objective of verifying if the low and high levels of the expression of bimodal genes correlate to the underlying biological processes, associated to a better or worse prognosis of the disease. With the clinical variables of the patients, we draw profiles related to the relative levels of expression.