MicrobioLink: An Integrated Computational Pipeline to Infer Functional Effects of Microbiome-Host Interactions

Abstract

Microbiome-host interactions play significant roles in health and in various diseases including autoimmune disorders. Uncovering these inter-kingdom cross-talks propels our understanding of disease pathogenesis and provides useful leads on potential therapeutic targets. Despite the biological significance of microbe-host interactions, there is a big gap in understanding the downstream effects of these interactions on host processes. Computational methods are expected to fill this gap by generating, integrating, and prioritizing predictions-as experimental detection remains challenging due to feasibility issues. Here, we present MicrobioLink, a computational pipeline to integrate predicted interactions between microbial and host proteins together with host molecular networks. Using the concept of network diffusion, MicrobioLink can analyse how microbial proteins in a certain context are influencing cellular processes by modulating gene or protein expression. We demonstrated the applicability of the pipeline using a case study. We used gut metaproteomic data from Crohn's disease patients and healthy controls to uncover the mechanisms by which the microbial proteins can modulate host genes which belong to biological processes implicated in disease pathogenesis. MicrobioLink, which is agnostic of the microbial protein sources (bacterial, viral, etc.), is freely available on GitHub.

Keywords: computational pipeline; microbiota–host interactions; network diffusion; networks; protein–protein interactions; systems biology.

Figure 1

Graphical representation of the MicrobioLink workflow.

Figure 2

(A) Graphical representation of the signaling paths between curated receptor proteins (host proteins predicted to be modulated by microbial proteins) and target autophagy genes: The network compilation was performed by tracing the signaling chains from the human receptors (predicted to interact with the bacterial proteins) to the autophagy genes using the TieDIE tool which adopts a network diffusion [51]. For brevity, only the results corresponding to the domain–motif interaction analysis are discussed in the case study. (B) Network representing the signalling chains after exclusion of proteins connected with bacterial proteins detected in both CD and healthy conditions: Proteins present in both conditions that were retained were those directly regulating the target autophagy genes. (C) Network obtained by retaining only chains with transcriptional regulatory interactions between the intermediary protein (3rd layer) and the target autophagy genes (4th layer): The immediate upstream proteins from the autophagy target genes were confined to transcription factors modulating the target autophagy genes via a transcriptional regulatory interaction.

Figure 3

Final network model consisting of the proteins and the biological processes in which they are involved, inferred from the Gene Ontology (GO) enrichment test.