Network entropy metrics to assess keystone taxa influence in microbiome-host interaction under antibiotic-driven selective pressure

Understanding the dynamics of microbiome-host interactions under antibiotic-driven selective pressure is critical for addressing microbial dysbiosis and ensuring host health. While compositional and functional analyses have advanced our knowledge of microbial ecosystems, they often fall short in identifying the structural roles of specific microbial taxa. Keystone taxa species that disproportionately influence microbiome stability and host physiology are pivotal in maintaining ecological balance. However, quantifying their systemic influence remains challenging, especially under perturbations such as antibiotic exposure. This study presents a novel framework leveraging network entropy metrics to evaluate the influence of keystone taxa in microbial co-occurrence networks during antibiotic treatment. By applying information-theoretic measures such as Shannon entropy, local node entropy, and global network entropy, we capture how antibiotic-induced disruption alters microbial interaction complexity and the centrality of key taxa. Using longitudinal 16S rRNA sequencing data from murine gut microbiomes subjected to broad-spectrum antibiotics, we reconstructed dynamic interaction networks and tracked entropy changes over time. Our findings reveal that antibiotics reduce overall network entropy, indicating collapse in microbial diversity and connectivity. More importantly, taxa with high local entropy shifts particularly Bacteroides and Lactobacillus species demonstrated outsized influence on network restructuring and host inflammatory markers. These results suggest that entropy-based metrics can identify functional keystone species beyond mere abundance or frequency. This entropy-driven framework offers a scalable, quantitative approach to evaluate microbial resilience, optimize probiotic interventions, and inform precision antimicrobial therapies. Future research can integrate multi-omic and spatial data to refine these insights across different host niches and microbial ecosystems.