Many illnesses linked to bacteria, such as inflammatory bowel disease and colorectal cancer, are often connected to an overabundance of certain gut bacteria that are thought to be harmful. However, researchers, utilizing a machine learning algorithm, found that changes in the gut’s microbial density — known as microbial load — might contribute more to the presence of bacteria associated with diseases than the diseases themselves.
Many illnesses linked to bacteria, such as inflammatory bowel disease and colorectal cancer, are often connected to an overabundance of certain gut bacteria that are thought to be harmful. However, researchers, utilizing a machine learning algorithm, found that changes in the gut’s microbial density — known as microbial load — might contribute more to the presence of bacteria associated with diseases than the diseases themselves.
In a report published on November 13, 2024, in the Cell Press journal Cell, the researchers highlighted that variations in a patient’s microbial load are influenced by various factors, including age, sex, diet, country of origin, and antibiotic use. This was identified as a significant factor impacting bacterial profiles in fecal samples of both healthy individuals and those with diseases.
“We were taken aback to discover that many microbial species, previously thought to directly relate to diseases, were instead more closely linked to fluctuations in microbial load,” states Peer Bork from the European Molecular Biology Laboratory (EMBL) Heidelberg, one of the study’s lead authors. “This suggests that these species are primarily associated with symptoms like diarrhea and constipation rather than being directly tied to the diseases themselves.”
Microbial load has been acknowledged as crucial in microbiome studies, but extensive research has been limited due to costly and labor-intensive methods. The researchers overcame this challenge through a machine learning approach, developing a prediction model for fecal microbial load based on relative microbiome composition and applying it to large metagenomic datasets to investigate variations in both health and disease.
“Determining microbial load from fecal samples requires considerable effort, and we were fortunate to have access to two large metagenomic datasets where this load had been accurately measured,” notes Michael Kuhn, another senior author from EMBL. “Our goal is to generalize these findings for the benefit of the broader research community, allowing microbial load to be predicted for all studies concerning the adult human gut microbiome.”
The datasets created during this research consist of thousands of metagenomes and experimentally assessed microbial loads from the EU-funded GALAXY (Gut-and-Liver Axis in Alcoholic Liver Fibrosis) project and the Novo Nordisk Foundation’s MicrobLiver project. They also utilized data from the public MetaCardis study population and included exploratory datasets from tens of thousands of previous studies, incorporating populations from Japan and Estonia.
However, the researchers acknowledge certain limitations in their work. Since their analysis was based purely on associations, they could not definitively establish causality or provide insights into the underlying mechanisms. Moreover, the method they developed is specific to the human gut microbiome; different training datasets would be necessary to forecast microbial load in other environments.
Looking ahead, the research team intends to focus on microbial species that are more directly linked to diseases, regardless of microbial load, to deepen understanding of their roles in disease development and evaluate their potential as biomarkers. Furthermore, adapting this prediction model for other ecosystems, such as ocean and soil microbiomes, could offer additional insights into microbial ecology on a global scale.
