A resource for the comparison and integration of heterogeneous microbiome networks

Abstract

Naturally occurring microbial communities often comprise thousands of taxa involved in complex networks of interactions. These interactions can be mediated by several mechanisms, including the competition for resources, the exchange of signals and nutrients, cell-cell contact and antibiotic warfare. In addition to direct measurements and computational predictions of interactions, abundant data on microbial co-occurrence associations can be inferred from correlations of taxa across samples, which can be estimated from metagenomic, and amplicon datasets. The analysis and interpretation of interaction and correlation networks are limited by the challenge of comparing across different datasets, due to heterogeneity of the data itself and to the lack of a platform to facilitate such comparisons. Here, we introduce the Microbial Interaction Network Database (MIND) - a web-based platform for the integrative analysis of different types of microbial networks, freely available at http://microbialnet.org/. In addition to containing a growing body of curated data, including amplicon-based co-occurrence networks, genome-scale model-derived networks, metabolic influence networks and horizontal gene transfer networks, MIND allows users to upload and analyze newly generated networks using a JSON format and standard NCBI taxonomy. The platform provides convenient functions to compare and query multiple networks simultaneously, and to visualize and export networks and datasets. Through some illustrative examples, we demonstrate how the platform might facilitate discoveries and help generate new hypotheses on host-associated and environmentally important microbial ecosystems through the power of knowledge integration.