Hypothesis testing phase

WARWICK DUNN: So we have performed our discovery study, identified the statistically significant metabolite, or metabolites, and constructed a hypothesis. Well, what is next? Well, the process of discovery is only the first step of the scientific study. We now need to validate the discovery and test the hypothesis we have constructed. This stage is the hypothesis testing phase. So what do you need in the hypothesis testing experiment? How we design this second study to test a hypothesis is hugely dependent on the biological context. For example, what do we do if we have identified a metabolic reaction that is important in the biological mechanism? Well, we can test this hypothesis biologically by perturbing the system.

The perturbation could be to knock out, or enhance the metabolic reaction and measure changes in the phenotype. We could also measure changes in other components of the biological pathway at different functional levels. For example, investigate whether a change in a metabolic reaction is a consequence of changes at the proteomic level, including post-translational modification, or changes in enzyme activity. A different scenario would be when we have identified a single metabolite, or group of metabolites present in the biofluid, which act to predict your risk of developing the disease in five years’ time. We would look to test the hypothesis in an independent sample set. That is a set of biofluid samples that are different to those used in your discovery phase experiment.

We would also need to develop and validate an analytical method to accurately and precisely measure the concentration of the metabolites of interest. Because we know which metabolites to look for, we can apply a targeted analytical method to detect only the metabolites of interest. The analytical method should be precise and accurate in identifying the list of metabolites in the appropriate biological matrix. So if we’re using a urine sample to detect the metabolites we need to check that the background matrix in urine does not interfere in the targeted assay. The method also needs to be sufficiently sensitive to detect the metabolites in the sample. And we will discuss the analytical approaches that we apply in week three of the course.

In the validation study, the metabolites of interest are known. If we use our jigsaw analogy from earlier, these metabolites analysed in the targeted approach represent the most important jigsaw pieces and show us the key image present in the jigsaw picture. One validation step may not be sufficient to fully validate the results. The route from discovery to validation may require several metabolimic, biological, or clinical studies. The level of validation will depend on the environment and biological, or clinical context. For example, the validation of a biomarker to apply in clinical practise will require a greater level of validation and will take longer to complete than validation of a biological mechanism in yeast.

For biomarkers to be applied in clinical practise after the validation stage is performed, the outcome of your study can be translated into routine practise operating to specific regulatory requirements and standards. This is called the translation phase. The validation and translation phases can take many years to complete. This is the final output of your discovery phase experiment. The translation phase is very important. It’s the output of our research to benefit human population and provides the impact of our research.