Résumé
With the increasing number of genomes being sequenced, a major objective is to transfer accurate annotation from characterised proteins to uncharacterised sequences. Consequently, comparative genomics has become a usual and efficient strategy in functional genomics. The release of various annotated genomes of plants, such as O. sativa and A. thaliana , has allowed setting up comprehensive lists of gene families defined by automated methods. However, like for gene sequence, manual curation of gene families is an important requirement that has to be undertaken. GreenPhylDB comprises protein sequences of 12 plant species fully sequenced that were grouped into homeomorphic families using similarity-based methods. Clusters are finally processed by phylogenetic analysis to infer orthologs and paralogs that will be particularly helpful to study genome evolution. Previously, each cluster has to be curated (i.e. properly named and classified) using different sources of information. A web interface for plant gene families’ curation was developed for that purpose. This interface, accessible on GreenPhylDB ("http://greenphyl.cirad.fr":http://greenphyl.cirad.fr), centralizes external references (e.g. InterPro, KEGG, Swiss-Prot, PIRSF, Pubmed) related to all gene members of the clusters and shows statistics and automatic analysis. We believe that this synthetic view of data available for a gene cluster, combined with basic guidelines, is an efficient way to provide reliable method for gene family annotations.