| Abstrakt: |
Considering biological systems as information processing entities and analyzing their organizational structure via information-theoretic measures has become an established approach in life sciences. We transfer this framework to a field of broad general interest, the human gut microbiome. We use BacArena, a software combining agent-based modelling and flux-balance analysis, to simulate a simplified human intestinal microbiome (SIHUMI). In a first step, we derive information theoretic measures from the simulated abundance data, and, in a second step, relate them to the metabolic processes underlying the abundance data. Our study provides further evidence on the role of active information storage as an indicator of unexpected structural change in the observed system. Besides, we show that information transfer reflects coherent behavior in the microbial community, both as a reaction to environmental changes and as a result of direct effective interaction. In this sense, purely abundance-based information theoretic measures can provide meaningful insight on metabolic interactions within bacterial communities. Furthermore, we shed light on the important however little noticed technical aspect of distinguishing immediate and delayed effects in the interpretation of local information theoretical measures. Author summary: The idea of considering biological systems as information processing systems has recently gained increasing attention in life sciences. Following this approach, one assumes that every agent of a living system stores and transfers information and that those operations reflect biological processes. Consequently, one should be able to gain knowledge on a system's biological processes by measuring its information processing. An established mathematical framework to quantify information processing based on observational data is given by information theory. We explore this idea in the context of bacterial community analysis in the human gut habitat. This complex system is a common subject of ongoing research due to its impact on immunity and health. To keep complexity manageable and allow for controlled intervention, we simulate metabolic interaction of seven key species, representing the essential metabolic potential of the human gut microbiome. However, our approach can likewise be directly applied to in vivo/in vitro data. Estimating information storage and transfer from the resulting abundance data, we first identify a technical artefact, whose disregard can lead to strong misinterpretations of information processing. Second, making use of the metabolic data that the simulation provides, we show that information storage and transfer indeed relate to environmental conditions and species' interactions in a meaningful way. We derive general principals for the interpretation of these measures, that allow us to infer patterns of bacterial interaction without the need to measure chemical processes. In this sense, our approach could provide an alternative way to study impacts of dietary plans or medical interventions on the community level. [ABSTRACT FROM AUTHOR] |
