A survey of computational approaches for characterizing microbial interactions in microbial mats

Abstract

In this review, we use microbial mat communities as a general model system to highlight the strengths and limitations of current computational methods for analyzing interactions between members of microbial ecosystems. We describe the factors that make this environment have such a high degree of interaction, and we explore different categories of both laboratory and computational tools for studying these interactions. For each tool, we describe efforts to apply them to microbial mats in the past and, in the process, argue that genome-scale metabolic models have breakthrough potential for modeling microbial interactions in microbial mats.

Keywords: Comparative genomics; Computational models; Environmental data; Meta-omics; Metabolic models; Microbial ecology; Microbial evolution; Microbial mats; Transdisciplinary studies.

Fig. 1

Multi- and interdisciplinary approach to the study of microbial mats. The complexity of microbial mats can not be approached by a single discipline and requires multiple scientists from different fields to understand them

Fig. 2

Relationships between biotic and environmental factors that influence microbial mat communities and behaviors. Red and black arrows in the figure highlight the influence of biotic and environmental factors over others, respectively. Dashed arrows show indirect influence between factors. See text for explanation

Fig. 3

Computational model approaches to study microbial interactions. Top-down and bottom-up models used nowadays with their respective inputs. Within top-down models, network inference models, and ecological ones use abundance matrices to predict community dynamics, while HGT detection models use assembly graphs and assembled contigs to identify structural changes that appear out of place, with some methods particularly focused on those representing HGT events. Bottom-up models use either stoichiometric matrices or metabolite concentrations and kinetic and thermodynamic constants to infer microbial interactions. The results of these two types of models predict community dynamics and microbe-microbe interactions, although bottom-up models can also predict far more granular outcomes