Microbiomes are dynamic, forming a “social network” interacting with each other and their host through different metabolites. The host physiology, age, or gender can confound the biological processes and the microbiota. Thus, longitudinal data are necessary to understand the complex interaction. Metatranscriptomics, metagenomics, and metabolomics data generated by the Integrative Human Microbiome Project that followed 132 individuals with inflammatory bowel disease over one year were used in our analyses. We compensated for the variable biological process, non-uniform sampling, noisy and missing data by using Dynamic Bayesian Networks (DBNs). The DBNs are ideally suited to model heterogeneous dynamic systems and infer temporal interactions between their constituents. Using DBN, a framework was developed to identify relationships between genes, taxa, and metabolites. MIMOSA validated a significant (based on a Poisson-Binomial distribution) number of the edges with the highest bootstrap confidence predicted by the DBN. In-depth knowledge of this network would improve our understanding of the disease, metabolic potential of each taxon, and potentially critical metabolites that may lead to better therapeutics. The work represents novel and valuable research on an integrated analysis of multiomic longitudinal data.