Colorectal cancer is the world’s second deadliest cancer. It’s treatable if caught early, but colonoscopies, the main way to detect it, are expensive and uncomfortable, so many people delay getting checked.
That delay often means the cancer is found late, when treatment options are fewer and less effective. This makes it urgent to develop easier, less invasive ways to diagnose it, especially since cases are mysteriously rising among young adults. Researchers know gut bacteria are linked to colorectal cancer, but applying this knowledge in real-world medicine is challenging. Why?
Because even bacteria from the same species can behave differently, some may fuel cancer, while others do nothing at all.
Researchers at the University of Geneva have pulled off something remarkable: using machine learning, they mapped every human gut bacterium down to its subspecies. Why does that matter? Because this ultra-detailed inventory lets researchers understand which microbial subgroups actually influence our health, and which ones might be waving red flags for diseases like colorectal cancer.
We all have different gut bacteria
The twist? Instead of relying on invasive colonoscopies, they used this bacterial map to detect cancer from simple stool samples.
Mirko Trajkovski, full professor in the Department of Cell Physiology and Metabolism and in the Diabetes Centre at the UNIGE Faculty of Medicine, who led this research said, “Instead of relying on the analysis of the various species composing the microbiota, which does not capture all meaningful differences, or of bacterial strains, which vary greatly from one individual to another, we focused on an intermediate level of the microbiota, the subspecies.”
“The subspecies resolution is specific and can capture the differences in how bacteria function and contribute to diseases, including cancer, while remaining general enough to detect these changes among different groups of individuals, populations, or countries.”
Matija Trickovic, a PhD student, helped create a smart new way to analyze vast amounts of gut bacteria data. His team built the first detailed list of all the tiny bacterial subgroups living in our gut, and figured out how to use that list for both science and medical tests.
By combining this bacteria catalogue with real patient data, they built a model that can spot colorectal cancer just by looking at the bacteria in a stool sample, without the need for invasive tests.
Their method found 90% of cancer cases, almost as accurate as colonoscopies (which detect 94%) and better than any other non-invasive test available today.
With more patient data, this tool could become even more precise. It might become a regular screening method, helping doctors catch cancer early and only use colonoscopies when truly needed.
Researchers are teaming up with Geneva University Hospitals to run a clinical trial. They aim to assess the effectiveness of their method in identifying various stages of colorectal cancer and the types of lesions it can detect.
By zooming in on tiny differences between bacterial subspecies, researchers are starting to understand how gut bacteria affect our health. That means this same technique could be used to create simple, non-invasive tests for many other diseases, all from one stool sample.
Journal Reference:
- Matija Tričković, Silas Kieser, Evgeny M. Zdobnov et al. Subspecies of the human gut microbiota carry implicit information for in-depth microbiome research. Cell Host & Microbe. DOI: 10.1016/j.chom.2025.07.015