Gut Bacteria Interactions, Not Just Species, Distinguish Healthy From Diseased Microbiomes

Ask most researchers what makes a gut microbiome healthy, and the answer typically involves species counts - how many bacterial types are present, which ones predominate, which are missing. This framing has guided a decade of microbiome research, producing long lists of bacteria associated with various conditions. It has also, in many cases, produced therapies that work inconsistently and predictions that do not hold across populations.

A team from Rutgers University, Universidad de Granada in Spain, and Princeton University started from a different premise. Instead of cataloguing which bacteria occupy the gut, they asked how those bacteria relate to one another. The distinction turns out to matter considerably.

From Computer Models to Patient Data

The project began computationally. Roberto Corral Lopez, then a Fulbright doctoral visiting student at Rutgers and now a postdoctoral researcher at the Instituto Carlos I de Fisica Teorica y Computacional in Spain, built computer simulations of how gut bacteria compete for nutrients and exchange metabolic byproducts. The goal was modest: test whether the models could reproduce basic features of real microbiomes.

Early results surprised the team. The simulations naturally settled into two distinct configurations - a diverse, competitive state and a second state dominated by small, tightly connected groups of cooperating bacteria. When the researchers compared these patterns against DNA sequencing data from actual patients, the same two configurations appeared.

That observation prompted the development of a formal measurement tool. The team constructed what they call the Ecological Network Balance Index, or ENBI, which captures whether a microbial community is dominated by competitive or cooperative interactions. Applied to existing datasets from patients with a range of gut diseases, ENBI consistently separated healthy individuals from patients - not by identifying specific bacteria but by characterizing the relational structure of the whole community.

In colorectal cancer, the index did something additional: it rose as the disease progressed. This suggests ENBI may function not just as a binary classifier but as a continuous measure that tracks disease severity.

Why Fecal Transplants Sometimes Work and Sometimes Do Not

The ecological framing also offers a new interpretation of why some gut therapies produce inconsistent results. Probiotics, for instance, are typically designed around the concept that specific beneficial bacteria need to be introduced. If the real issue is not which bacteria are present but how they interact, adding a single species into a dysfunctional community may accomplish nothing - the new bacterium arrives in an environment where the cooperative-competitive balance is already wrong.

Fecal microbiota transplants, which transfer entire microbial communities from a donor, might work precisely because they transplant the relational network, not just a collection of species. As Juan Bonachela, associate professor in the Department of Ecology, Evolution and Natural Resources at Rutgers, explained, the benefit may come from restoring the interactions that allow a healthy community to function - not from introducing individual species per se.

That hypothesis has implications for donor selection. Current practice in fecal transplants relies primarily on basic health screening and availability. If interaction networks matter as much as species composition, matching donor communities based on how their microbial networks are structured might produce more predictable outcomes than the current approach.

What the Measurement Does Not Yet Resolve

ENBI is currently a research tool applied to existing datasets, not a clinical instrument. Several steps separate demonstrating that an index distinguishes healthy from diseased samples in retrospective data from using that index to monitor individual patients or guide treatment decisions.

The study identifies colorectal cancer, inflammatory bowel disease, C. difficile infection, and irritable bowel syndrome as conditions where ENBI separated patients from healthy controls. But the underlying datasets used for this analysis were collected for other purposes and may not represent the full clinical diversity of each condition. Prospective studies - following patients over time and measuring ENBI alongside clinical outcomes - are the logical next step for evaluating the index's practical utility.

In principle, ENBI could be calculated from stool samples, making it a non-invasive monitoring approach. Whether it outperforms existing diagnostic markers, and in which specific contexts, remains to be shown. The research was published in Science.