Integration of metataxonomic datasets into microbial association networks highlights shared bacterial community dynamics in fermented vegetables

The management of food fermentation is still largely based on empirical knowledge, as the dynamics of microbial communities and the underlying metabolic networks that produce safe and nutritious products remain beyond our understanding. Although these closed ecosystems contain relatively few taxa, they have not yet been thoroughly characterized with respect to how their microbial communities interact and dynamically evolve. However, with the increased availability of metataxonomic datasets on different fermented vegetables, it is now possible to gain a comprehensive understanding of the microbial relationships that structure plant fermentation. In this study, we present a bioinformatics approach that integrates public metataxonomic 16S datasets targeting fermented vegetables. Specifically, we developed a method for exploring, comparing, and combining public 16S datasets in order to perform meta-analyses of microbiota. The workflow includes steps for searching and selecting public time-series datasets and constructing association networks of amplicon sequence variants (ASVs) based on co-abundance metrics. Networks for individual datasets are then integrated into a core network of significant associations. Microbial communities are identified based on the comparison and clustering of ASV networks using the "stochastic block model" method. When we applied this method to 10 public datasets (including a total of 931 samples), we found that it was able to shed light on the dynamics of vegetable fermentation by characterizing the processes of community succession among different bacterial assemblages.