Predicting the structure of microbial communities

Our body is home to a large number of microbes. Our health depends on them. For instance, disruption of the microbial communities in our gut, which can happen with the use of antibiotics, affects not only our digestion but also our mental health. Microbial communities are also found in nearly every environmental niche, including the soil and the oceans, and are central players in the respective ecosystems. More recently, recognizing their power, artificial microbial communities are being assembled for numerous applications, such as biofuel production. It is important, therefore, to understand how microbial communities thrive and devise ways of engineering them.

In a recent study, Prof. Narendra Dixit together with his M. Tech. student Aamir Ansari developed a new method to efficiently predict the compositions of microbial communities. The study was performed in collaboration with Unilever R&D in Bengaluru. The method, termed EPICS, involves the use of effective pairwise interactions to predict community structures.

A major challenge in such prediction was the large number of interactions between the microbes present, which determine the compositions of microbial communities but remain difficult to unravel. In the new method, the researchers developed a way to subsume higher order interactions into effective pairwise interactions, which were easier to unravel and, at the same time, facilitated accurate prediction of community compositions. The method dramatically reduces the effort in unravelling the interactions and enables the study of much larger microbial communities than currently feasible. They demonstrated its applicability using a synthetic oral microbiome. The method brings us a step closer to understanding the microbiomes pervading our bodies and surroundings and developing microbiome-based interventions.

Ansari AF, Reddy YBS, Raut J, Dixit NM. An efficient and scalable top-down method for predicting structures of microbial communities. Nature Computational Science 1, 619-628 (2021).