Discovery based studies in metabolomics

MARK VIANT: Metabolomics is highly amenable to high-throughput studies. Recent advances in scientific technologies, computer power, and data analysis techniques have provided the first platform to perform these hypothesis-generating studies. These are also called discovery based studies. Nowadays, in a single study we can simultaneously measure hundreds or even thousands of metabolites present in literally thousands of different samples, and then interrogate the data with powerful data analysis techniques to acquire new biological insights. Hypotheses-generating studies have evolved over the last 15 years within the so-called omics scientific fields and are the primary route to yielding new biological discoveries. By performing hypothesis-generating studies to measure a wide diversity of metabolites– hundreds or even thousands of them– we can investigate global changes in the metabolome.

In contrast, before the development of metabolomics, traditional hypothesis-testing approaches were limited to studying just a few metabolites.

In hypothesis-generating studies the objectives are either to determine a metabolite– or, more likely, to determine a group of metabolites– that change as a result of a perturbation to a biological system. This perturbation might, for example, be caused by an environmental pollutant. Or it might be a virus that causes a human disease. These studies help us to identify targets for drug development. Also, to monitor the health of humans and organisms living in the environment. And to discover the molecular mechanisms that underpin biological processes. These studies can only be achieved if a wide variety of metabolites are measured.

This is because we need to investigate all of the molecular parts of a biological system in order to understand how that complete system works. As an example, think of a jigsaw. If you have a few pieces of a 500-piece jigsaw puzzle, you obviously cannot visualise the full picture. It’s only when you have all, or at least most, of those 500 pieces together that you can then visualise the full picture. Hypothesis-generating studies operate in the same way. Metabolomics helps us to visualise all of the metabolic jigsaw pieces. Metabolites are not synthesised in isolation, and they do not operate in isolation. They synthesise and operate in what’s called a metabolic network.

While traditional diagnosis techniques may have focused on a specific biomarker to diagnose a disease, such as blood glucose to diagnose diabetes, it’s now common to assess multiple factors when diagnosing the risk of a disease. For example, when assessing someone’s risk of heart disease, we look at their age, body mass index or BMI, their family history, cholesterol and triglyceride levels. In the same way, studying different characteristics to define the risk of developing a disease, by measuring hundreds to thousands of metabolites, we can identify those biologically important metabolites. A hypothesis-generating study will often be the first metabolomics study performed, which is then followed up by one or more hypothesis-testing studies to follow up and validate the discoveries that you’ve made using metabolomics.

And the final step of the study will be to translate the discovery to impact positively on human or environmental health. In hypothesis-generating studies, the data is acquired in an untargeted approach, as described earlier, to maximise the amount of information collected. And then use a combination of univariate and multivariate statistical analysis methods to interrogate that data, and determine which metabolic changes occur between the different experimental conditions. We will discuss how this data analysis is done in week four of the course. To translate the discoveries from metabolomics studies to their real-world application, there must be further biological experiments conducted to test and validate the hypotheses constructed from the discovery studies. And we will discuss this in the next section.

The key message from performing any metabolomics study is that the discovery phase experiment is only the first step, and the project is not complete without the validation steps to confirm your findings, and then to translate those findings into the relevant working environment.