Phytic acid is an antinutritional factor in cereal feeds, and the use of phytases increases the bioavailability of nutrients bound to this molecule. However, the application of these enzymes depends on their thermal stability and activity at acidic pH. Therefore, in this study we created a database composed of 59 phytase sequences and analyzed the interactions that stabilize their structures in order to understand whether they contribute to the biochemical properties observed. The sequences were aligned and grouped at 30 % similarity, generating 5 clusters, which highlights the high variability among them. A comparative structural analysis of the cluster 3 phytases revealed conserved catalytic domains, as well as eight cysteine residues along the primary sequence, forming disulfide bonds for stabilizing the three-dimensional structure. However, the number of Van der Waals, ionic, and hydrogen interactions, and disulfide bonds was not determinant for the biochemical characteristics presented by these enzymes. The phytase KM873028, from cluster 3, was selected for characterization studies, but its expression in Pichia pastoris generated a protein with properties distinct from those derived from the same sequence expressed in a prokaryotic system. It is likely that the differences observed are associated with the location of the interactions in the structures, non-conserved amino acid residues found around the catalytic site, and post-translational modifications inherent to the expression systems. These possibilities highlight the relevance of strategic choices related to enzyme expression aiming at its production and industrial feasibility.