A Novel Computational Approach for the Mining of Signature Pathways Using Species Co-occurrence Networks in Gut Microbiomes

Background: Advances in metagenome sequencing data continue to enable new methods for analyzing biological systems. When handling microbial profile data, metagenome sequencing has proven to be far more comprehensive than traditional methods such as 16s rRNA data, which rely on partial sequences. Microbial community profiling can be used to obtain key biological insights that pave the way for more accurate understanding of complex systems that are critical for advancing biomedical research and healthcare. However, such attempts have mostly used partial or incomplete data to accurately capture those associations.

Methods: This study introduces a novel computational approach for the identification of co-occurring microbial communities using the abundance and functional roles of species-level microbiome data. The proposed approach is then used to identify signature pathways associated with inflammatory bowel disease (IBD). Furthermore, we developed a computational pipeline to identify microbial species co-occurrences from metagenome data at various granularity levels.

Results: When comparing the IBD group to a control group, we show that certain co-occurring communities of species are enriched for potential pathways. We also show that the identified co-occurring microbial species operate as a community to facilitate pathway enrichment.

Conclusions: The obtained findings suggest that the proposed network model, along with the computational pipeline, provide a valuable analytical tool to analyze complex biological systems and extract pathway signatures that can be used to diagnose certain health conditions.