In biotechnology, anaerobic digestors are commonly inoculated with an empirically chosen seed community. Using this procedure, one easily oversees the vast functional potential of less obvious microbial communities from natural or engineered ecosystems. We propose a systematic strategy to assess ecosystem function for an array of methanogenic inocula and related the performance of the microbial communities to their respective metagenomes. The success of our work heavily depends on a multidisciplinary effort shared by microbial ecologists and mathematicians. In a sequence of three experimental steps, we will (1) relate ecosystem performance to signatures of biodiversity derived from the metagenome structure. These signatures are used to (2) reverse-model a most probable virtual ancestor community. Finally, using the results form step 2, it may become possible to (3) predict ecosystem performance from an unknown inoculum. Our generic approach can be applied to a variety of ecosystems of interest to INRA. It will yield results with high possible impact for biotechnology and community ecology.