A multi-omics strategy for the study of microbial metabolism: application to the human skin's microbiome

The revolution of ‘omics’ technologies highlighted that associated microorganisms (also called microbiota) are integrated into the metabolic functions of their hosts. Yet when performing any particular type of ‘omics’ experiment, be it metabolomics, transcriptomics, or (meta)genomics, it is extremely difficult to interpret the observed relationships between metabolites, transcripts, and microbial species. This is due to the massive amount of data generated for each ‘omics’ technology, but also the cognitive challenge of interconnecting these observations and contextualizing them in their biological (eco)system. For these reasons, there is a need for testing methods that can facilitate the translation of these ‘omics’ experimental observations into putative molecular processes or biological interactions. To accelerate the interpretation of ‘omics’ data from a description of microbial, transcript, or metabolite identities or abundances into a functional understanding of the interplay between the individual entities of the biological system, we designed a novel multi-omics strategy for the annotation and integration of metabolomics and metagenomics data. We generated metabolome and microbiome datasets by LC-MS/MS based metabolomics profiling and metagenomic sequencing, respectively. Comprehensive metabolite annotations were obtained by molecular networking and computational annotation of fragmentation spectra. Associations between microbes and GNPS molecular networks were predicted by machine learning and visualized as an extensively annotated, nested interaction network in Cytoscape. As a proof of concept, we applied this strategy to scalp swabs from a cohort of healthy volunteers with varying scalp sebum levels and were able to elucidate the antagonistic interaction between two well-characterized microbes,Staphylococcus epidermidisandCutibacterium acnes.