What is already known
For years, scientists have focused on the microbial composition of the human gut. However, examining the relative abundance of different species is not always enough to identify patterns linked to diseases. Unlike microbial composition, microbial load measures the total number of microbial cells in the gut, providing a useful metric for understanding the microbiota and its links to disease.
What this research adds
Researchers developed a machine-learning model that is able to predict microbial load using only metagenomic data — the genetic material collected directly from the human gut, without the need for expensive and time-consuming lab tests. By applying this model to nearly 28,000 samples from large studies, the team found that microbial load is a major determinant of microbiota variation and plays a key role in disease associations.
Conclusions
The findings suggest that including microbial load in microbiota studies can help researchers better understand how gut bacteria influence diseases, leading to more accurate diagnoses and effective treatments.
The microbial communities in people’s guts differ in both composition and total abundance, influencing their functions and interactions within the body. Researchers have now developed a machine-learning model capable of predicting microbial load, or the total number of specific microbes in the gut, uncovering it as a key driver of microbiota variation and its links to diseases.
The findings, published in Cell, suggest that including microbial load in microbiota studies can help researchers better understand how gut bacteria influence diseases, leading to more accurate diagnoses and effective treatments.
The work may also have implications beyond the gut microbiota. “Our oceans, soils, rivers – are all teeming with microbes, and understanding these microbiomes could yield valuable insights to help preserve our planetary health,” says study senior author Peer Bork at the European Molecular Biology Laboratory (EMBL) in Heidelberg, Germany. “This study shows us that microbial load is an important measure that must be taken into account in such studies. Thus, we will work towards translating the knowledge on the gut microbiome to other habitats.”
For years, scientists have focused on the microbial composition of the human gut. However, examining the relative abundance of different species is not always enough to identify patterns linked to diseases. Unlike microbial composition, microbial load measures the total number of microbial cells in the gut, providing a useful metric for understanding the microbiota and its association with health conditions.
So, Bork and his team set out to develop a machine-learning model that is able to predict microbial load using only metagenomic data — the genetic material collected directly from the human gut.
Predicting microbial load
To develop the AI-powered model, the researchers trained it on large datasets that contained both microbial composition, or the relative abundances of species, and experimentally measured microbial load, or the absolute number of microbial cells. Eventually, the model learned to predict microbial load using only the relative species abundances as input.
Next, the researchers validated the model on a dataset it had not encountered before. Then, they applied the model to a much larger dataset of nearly 28,000 samples collected from 159 studies across 45 countries. This analysis revealed that changes in microbial load are linked to health, diet, drugs and disease.
The team found that several factors affect microbial load. For example, diarrhea tends to decrease the number of microbes in the gut, while constipation can lead to an increase. On average, young people have a lower microbial load compared to older adults, and women have a higher microbial load than men — likely due to the higher frequency of constipation in women.
Open-access approach
Many microbial species that had been previously associated with disease were strongly linked to variations in microbial load, the researchers also found. “These findings suggest that changes in microbial load, rather than the disease itself, may be the driver of shifts in the microbiome in patients” says study lead author Suguru Nishijima at EMBL Heidelberg. “However, certain disease-microbe associations remained, and this shows that these are truly robust.”
The findings highlight the need to incorporate microbial load in microbiota studies to prevent false results, the authors say. Their model is openly accessible to researchers around the world for testing and reusing.
By predicting microbial load based only on metagenomics data, the authors add, this approach eliminates the need for expensive and time-consuming lab tests, and it could make microbial load analysis more accessible and scalable in the future.
