Testes estatísticos de classificação de domínios de proteínas em famílias e clãs

Abstract

The decisions of expert biologists on the recognition of Families and Clans of Protein Domains are usually confirmed by the application of algorithms of Dynamical Programming and techniques of Hidden Markov Chains in the literature of Computational Biology. The present work is an introduction to alternative statistical methods associated with the distribution of values of functions (Entropy Measures) of random variables (probabilities of occurrence of amino acids on the protein domains). The Hypothesis testing of ANOVA Statistics (one way) is then able to display the efficiency of these methods on a new confirmation of the existence of Families and Clans previously introduced.