Inflammatory bowel diseases (IBD) can be broadly divided into Crohn's disease (CD) and ulcerative colitis (UC) from their clinical phenotypes. Over 150 host susceptibility genes have been described, although most overlap between CD, UC and their subtypes, and they do not adequately account for the overall incidence or the highly variable severity of disease. Replicating key findings between two long-term IBD cohorts we have defined distinct networks of taxa associations within intestinal biopsies of CD and UC patients. Disturbances in an association network containing taxa of the Lachnospiraceae and Ruminococcaceae families, typically producing short chain fatty acids, characterize frequently relapsing disease and poor responses to treatment with anti-TNF-α therapeutic antibodies. Alterations of taxa within this network also characterize risk of later disease recurrence of patients in remission after the active inflamed segment of CD has been surgically removed.

At genome scale, it is not yet possible to devise detailed kinetic models for metabolism because data on the in vivo biochemistry are too sparse. Predictive large-scale models for metabolism most commonly use the constraint-based framework, in which network structures constrain possible metabolic phenotypes at steady state. However, these models commonly leave many possibilities open, making them less predictive than desired. With increasingly available –omics data, it is appealing to increase the predictive power of constraint-based models (CBMs) through data integration. Many corresponding methods have been developed, but data integration is still a challenge and existing methods perform less well than expected. Here, we review main approaches for the integration of different types of –omics data into CBMs focussing on the methods’ assumptions and limitations. We argue that key assumptions – often derived from single-enzyme kinetics – do not generally apply in the context of networks, thereby explaining current limitations. Emerging methods bridging CBMs and biochemical kinetics may allow for –omics data integration in a common framework to provide more accurate predictions.