The study of microbial communities and their interactions has attracted the interest of
the scientific community, because of their potential for applications in biotechnology,
ecology and medicine. The complexity of interspecies interactions, which are key for the
macroscopic behavior of microbial communities, cannot be studied easily experimentally.
For this reason, the modeling of microbial communities has begun to leverage the
knowledge of established constraint-based methods, which have long been used for
studying and analyzing the microbial metabolism of individual species based on
genome-scale metabolic reconstructions of microorganisms. A main problem of
genome-scale metabolic reconstructions is that they usually contain metabolic gaps due
to genome misannotations and unknown enzyme functions. This problem is traditionally
solved by using gap-filling algorithms that add biochemical reactions from external
databases to the metabolic reconstruction, in order to restore model growth. However,
gap-filling algorithms could evolve by taking into account metabolic interactions among
species that coexist in microbial communities. In this work, a gap-filling method that
resolves metabolic gaps at the community level was developed. The efficacy of the
algorithm was tested by analyzing its ability to resolve metabolic gaps on a synthetic
community of auxotrophic Escherichia coli strains. Subsequently, the algorithm was
applied to resolve metabolic gaps and predict metabolic interactions in a community of
Bifidobacterium adolescentis and Faecalibacterium prausnitzii, two species present in the
human gut microbiota, and in an experimentally studied community of Dehalobacter
and Bacteroidales species of the ACT-3 community. The community gap-filling method
can facilitate the improvement of metabolic models and the identification of metabolic
interactions that are difficult to identify experimentally in microbial communities.