ETD

• bioinformatics
• graph-theory
• metagenomics

The human gut houses a complex microbial ecosystem. Traditional metagenomic analyses of the gut microbiota rely solely on relative taxonomic abundance of each taxa and the analysis of diversity metrics. In order to have a more comprehensive overview on the role each taxa play in microbial community, independently of their abundance, a network-based framework was explored.
A low-biomass cohort of 535 neonatal meconium samples was used as a case study to construct a context-specific bipartite metabolic network. Given the low-biomass nature of meconium, the DECONTAM algorithm was implemented to pre-process the 16S rRNA gene sequencing data to distinguish true biological colonization from contamination. The network was created by mapping cleaned taxonomic profiles onto the literature-curated NJC19 large-scale metabolic network (Lim et al., 2020) while annotating those metabolic activities that are not described in the database. The network was filtered based on microbial presence in each sample and topological analyses were carried out to identify central nodes important for community stability. Finally, functional profiles inferred from the network and the ones predicted by the genome-driven algorithm Tax4Fun2, were compared. While the study’s framework represents a promising tool for analyzing metagenomic datasets, current intrinsic limitations must be addressed in future studies. Increasing taxonomic resolution and expanding scientific literature regarding metabolic information will be needed to further refine the accuracy of the network.