Enzymes have their function essentially determined by specific residue known as catalytic residue. Most methods to predict enzyme catalytic residue uses at least one criterion of primary sequence conservation. Considering that the function is a product of the structure, this project aims to identify among the structural descriptors available in the Sting software database those of higher relevance in the discrimination of catalytic residue. Thus, the project will propose a new enzyme classification method using such descriptors as input data. Expected results include: determination of specific signature descriptors for enzyme groups, constituted by consultations formed by inferior and superior values delimiting the intervals for each selected descriptor; creation of a "periodic table" of signatures of the protein' families and sub-families; multi-varied statistical data analysis of the Sting database with the intention of validating the method and identifying the new proteins' catalytic residue.