Onegevity Virtual Metabolomics
Onegevity Virtual Metabolomics
In the world of the human gut microbiome, the ability to measure the metabolic nature of the gastrointestinal ecosystem (human gut microbiome metabolome) is paramount to understanding its dynamic status and function. The gold standard for measuring these metabolites is through metabolomic methods, such as Nuclear Magnetic Resonance (NMR) and Mass Spectrometry, which are often costly and complicate the sample logistics process. However, there is a method that can elucidate this information simply, at low cost, and with high fidelity: Virtual Metabolomics.
Virtual Metabolomics allows the correlations between microbial DNA and microbial metabolic status to be modeled effectively. The technique utilizes data from large-scale initiatives (such as the Human Microbiome Project and the IBDMDB database) as well as data from peer-reviewed journal publications to infer the relative levels of molecules in the gut accurately without measuring those molecules directly. Virtual metabolomics is a central component of both Onegevity’s Gutbio and Performbio microbiome reports.
How Does Virtual Metabolomics Work?
There are a few distinct phases in the methodology Onegevity utilizes to model specific molecules in the microbiome using machine learning techniques. The first phase, training, uses data from large-cohort studies that include both human gut microbiome metagenomic (DNASeq) information and human gut metabolomic information taken from the population of individual trial participants. These data are used to discover the model that fits the best and predicts each metabolite with the highest accuracy. Once each model for each metabolite is optimized, then it can be used to make inferences. In the second phase, prediction, an incoming metagenomic sequence can be utilized as input for all optimized models that predict the specific metabolites – the output. These phases are described in more detail below.
Phase One: Training Phase
The objective of the training phase is to train the given data and learn from them in order to make accurate predictions.
- The initial process involves the collection of large-scale metagenomic data and metabolomic data from the same individual.
- Next, in the modeling and optimization step, the data are fit into a variety of machine learning algorithms in order to select the ones that describe the data with the highest accuracy.
- After data collection, modeling and optimization, each metabolite is assigned a prediction model, which is uploaded in Onegevity’s database. Every model can be then used to predict metabolite levels from metagenomic data.
The Training Phase is shown below in figure 1, The Prediction Phase methodology is shown in figure 2.
Figure 1. Methodology for training a model for metabolite prediction
Figure 2. Methodology for using a trained model to predict metabolite levels and scores
Phase Two: Prediction Phase
The objective of the prediction phase is to apply new metagenome samples to the models that were built and optimized in the training phase in order to infer the metabolite levels in these samples.
- The input for this process is a set of reads from the advanced Next-Generation metagenomic sequencing that Onegevity uses.
- The output is a set of scores that are reported on the Onegevity Gutbio and Performbio reports (see Gut Health section below for specifics).
Onegevity has checked for predictive reliability and clinical relevance before reporting metabolite levels in microbiome reports: amino acids (L-Valine, Taurine, and Creatine), B-vitamins (niacin - B3, B6, folate - B9, and B12), bile acids, and short-chain fatty acids (SCFA; butyrate, propionate, valerate, and organic acid lactate).
It is important to understand these metabolites levels contextually, meaning they are not absolute values, rather they are relative values compared to a population of individuals with similar characteristics. However, these levels have been shown to accurately predict the metabolic dynamics of the gut.
- The B vitamins play several different yet important roles in your body's functioning and are necessary for healthy neurological function and energy production. As a group, they aid in protein digestion, in the production of blood and immune cells, in the nervous system function, hormone production, and improvement of cholesterol levels. Assessing gut-derived B vitamin levels from a health-perspective is vitally important because of their significant production in the gut. Studies show that 86 percent of the recommended daily intake of vitamin B6 comes from the metabolic activity of gut bacteria in humans, 37 percent of folate (B9), 31 percent of vitamin B12, and 27 percent of niacin (B6).
- In the human body, SCFAs can: (1) act as an energy source and help our metabolism by improving blood lipid levels, increasing satiety, and improving sensitivity to insulin; (2) act as a signaling molecule in the nervous system, among other systems; (3) help prevent the absorption of toxic compounds and (4) increase nutrient circulation. In the gut, SCFAs from fibers can: (1) inhibit the growth of pathogens; (2) stimulate the growth of beneficial bacteria; (3) maintain a healthy pH, and (4) improve gut-immune capacity.
- Bile acids play an important role in the regulation of immune function within the gut. They are also significant effectors of the distributions of bacterial communities in the gut and contribute to motility. In the human body, bile acids have various roles including cholesterol metabolism, elimination of certain catabolites, and absorption of vitamins. Further, due to their similarity to steroid hormones, bile acids have functions that mimic the acid of hormones in certain cases.
- Amino acids are the building blocks of all proteins in the human body as well as of the molecular physiology of microbes. Due to their importance in both systems, amino acids from the diet as well as from the gut microbial metabolism are at equilibrium between the host (human) and the microbial (gut) systems.
Virtual Metabolomics is an advanced computational method that estimates accurately the metabolic characteristics of the gut microbiome. Through a process of training and prediction, Onegevity reports metabolite levels and insights based on large-scale databases, machine-learning optimized algorithms, and peer-reviewed publications. Scores that are representative of metabolite levels in a sample in comparison to the community can be further used as an accurate and cost-effective measure of the gut microbiome metabolic capability and gut health.