Many tasks in bioinformatics can be faced only using a combination of computational tools. In particular, functional annotation of gene products can be a very expensive task that may require the application of many analysis together with a manual intervention of biologists. In this area, the phylogenomics inference is one of the most accurate analysis methodologies for functional annotation that is not yet widely used due to the computational cost of some steps in its protocol. This paper discusses the implementation and deployment of such analysis protocol in a distributed grid environment using an agent architecture in order to simplify the interaction between users and the grid.