Csiszar's divergence measures are commonly applied in the analysis of genetic sequences. The establishing of segment size and threshold values can help to improve the entropic segmentation process. This paper describes the development of an optimization methodology for the entropic segmentation process, based in the discovering of the optimum segment size and threshold values. An iterative process was applied to the four serotypes of Dengue virus in order to identify the proteins position within each serotype. The obtained results showed that it is possible to establish parameters for discriminating the divergence measures values and also improve the sensitivity value when a protein is localized in the sequence.