The PICRUSt project aims to support prediction of the unobserved character states in a community of organisms from phylogenetic information about the organisms in that community. The primary application is to predict gene family abundance (e.g. the metagenome) in environmental DNA samples for which only marker gene (e.g. 16S rRNA gene) data are available. This is an open source, international, collaborative bioinformatics project developed in the Huttenhower, Beiko, Langille, Vega Thurber, Knight and Caporaso labs.