Scientists built a new machine-learning model to predict microbial load — the density of microorganisms in the guts — and used it to explain how microbial load plays an essential role in disease-microbiome relationships.
The billions of bacteria that live in the stomachs are lifelong companions, whether one is unwell or healthy. Scientists have demonstrated in recent decades how the characteristics of this “microbiome” might offer important hints about human illnesses and how to cure them.
According to a recent study from the Bork group at EMBL Heidelberg that was published in the journal Cell, the total number of microbes in the gut is influenced by a variety of factors, including lifestyle and illness. As a result, this often overlooked metric needs more analysis in the study of the gut microbiome.
From Composition To Loads
Researchers often concentrate more on microbial composition, or the proportion of various microbe species (mostly bacteria and archaea, but also protists, viruses, and other microorganisms), when examining microbiomes. This indicates, for instance, whether the amount of one species of bacteria in the guts of some disease patients increases or decreases in relation to other species.
Imagine that the gut contains only 1,000 microorganisms. In healthy individuals, this may include 10 bacteria of species ‘red’ and 20 bacteria of species ‘blue’, implying that red bacteria account for 2% of the microbiome and blue bacteria for 5%.
However, in individuals with a specific ailment, one can find that red bacteria account for 4% of the microbiome, indicating a relative rise, while blue bacteria remain at 5%. It is possible to hypothesize that the red bacteria are involved in this disease.
Microbial load, on the other hand, refers to the number of bacteria found in our stomachs. Experimentally, it is calculated as the number of microbial cells per gram of feces. Unlike microbiological composition, this is an absolute quantity. Consider the following scenario: disease reduces the total number of bacteria to 500. Looking at the absolute figures, it is plausible that the number of red bacteria remained constant while the number of blue bacteria fell.
Since the experimental techniques used to measure microbial loads are still time- and cost-intensive, scientists typically only take microbial composition into account when conducting microbiome studies.
Using Machine Learning to Make Microbiome Studies More Robust
We wanted to develop a new method that required no additional experimental methods to quantify microbial load. We had access to large datasets with both microbial composition and experimentally measured microbial load data. We wanted to see if we could use these to train a machine learning model to estimate microbial load given microbial composition alone.”
Suguru Nishijima, Study First Author, European Molecular Biology Laboratory
The GALAXY/MicrobLiver and Metacardis consortia, two large EU-funded initiatives to which the Bork Group has previously participated, provided the datasets used in this exercise. These data, which were collected from more than 3,700 people, offered the perfect means of determining if a machine-learning model could be trained to determine the overall count of microorganisms in a sample.
Indeed, Nishijima and his colleagues' algorithm was able to forecast microbial loads with a high degree of accuracy. They confirmed this by using a fresh dataset that the program had never experienced before. After confirming that the model was effective, the researchers used it on a sizable sample of more than 27,000 people, drawn from 159 earlier investigations carried out in 45 different countries.
They discovered that a wide range of factors can affect the microbial load. For instance, constipation can increase the number of bacteria in the gut, but diarrhea might decrease them. Younger persons have a lower average microbial load than older people, and women have a higher microbial load than males on average (perhaps related to the finding that women frequently suffer from constipation more than men). Microbial load is significantly altered by several diseases and the drugs used to treat them.
Nishijima added, “Importantly, many microbial species previously thought to be associated with disease were more strongly explained by variations in microbial load. These findings suggest that changes in microbial load, rather than the disease itself, may be the driver of shifts in the microbiome in patients. However, certain disease-microbe associations remained, and this shows that these are truly robust. This further confirms the importance of including microbial load in microbiome association studies to avoid false positives or false negatives.”
Future gut microbiome research can now take this crucial component into account because of the new machine learning model these researchers created, which is the first to forecast microbial loads from composition data. Researchers from all over the world are free to test and utilize the model.
Additionally, this might have effects that go well beyond the gut microbiome.
Our oceans, soils, rivers – are all teeming with microbes, and understanding these microbiomes could yield valuable insights to help preserve our planetary health. 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.”
Peer Bork, Study Senior Author, Group Leader, Director and Study Senior Author, European Molecular Biology Laboratory
A Closer Look at CAR T-Cell Therapy — Successes, Challenges, And Future