Research interests

The main interest of this lab is to explore microbial community structure and dynamics in silico and in vitro. We therefore work at the boundary of microbial ecology, systems biology and bioinformatics.



Network of microbial associations in the human dental plaque





Microbial association networks. The advent of metagenomics allows the inference of association networks from the presence/absence and abundance of microbial DNA. Two microbial taxa are considered to be associated if they co-occur more (positive association) or less (negative association) frequently across samples than expected by chance. Some of these associations may be due to ecological interactions such as mutualism, competition, commensalism or host-parasite relationships, thus highly associated microbial pairs are good candidates to look for an ecological relationship. In addition, the analysis of microbial association networks can give insights into the niche structure of microbial ecosystems (who lives together with whom), highlight taxa with many relationships and detect alternative communities.


Pathway predicted for the astCADBE operon in E. coli in a KEGG RPAIR network using NeAT tools.





Metabolic pathway prediction. During her PhD, Karoline Faust developed and evaluated methods to predict metabolic pathways from metabolic networks and a set of seed nodes (compounds/reactions). In contrast to pathway mapping or projection approaches, network-based pathway prediction can deal with variants and combinations of known pathways as well as predict novel pathways consisting of known reactions and compounds. Metabolic pathway prediction can be applied to assemble metabolic pathways from enzyme-coding genes that are assumed to be functionally related. These genes may be obtained from various sources: gene expression data, genomic organisation (operons, synteny), phylogenetic profiles, gene fusion data... Thus, pathway prediction can help to interpret gene expression data or to annotate bacterial genomes.