Measure twice, cut once: the importance of the design in AMR research

In the previous steps, we have glanced over the steps that are required to analyze resistomes. You will probably remember this image:

We will now go deeper into it, starting with Step 1.

The first obvious aspect of any study is to have a sound design of your experiment. This might seem simple but it is really the most crucial aspect of a research project. If the design is flawed, the resources to correct this during the analysis are limited. Often times, researchers are eager to get their hands-on work with the samples going fast and do not spend the time necessary at this stage. To avoid such scenario, researchers should be encouraged to think it through a number of times if the design allows for a comparison of the desired variables with the minimum bias possible. Often, the ‘perfect’ conditions are not attainable, so researchers come to the difficult decision of making compromises. One can only make this decision once the research questions being asked are very clear.

In terms of general factors in research design, resistome studies are similar to other types of research. Thus, as little variation as possible across comparison groups is wanted in order to reduce bias. And it is important to already incorporate these concepts in the design. For instance, as an example, let’s say you plan on collecting saliva samples from individual that received antibiotics to analyze the impact on the resistome. In this case, you would also want to collect control samples from individuals that did not receive the antibiotics and, instead, received a placebo drug. For recruiting purposes, it is important to try to obtain very similar populations to minimize bias. Another example, let’s say you collected all samples and now need to extract them, you would want to perform as many as possible in the same manner, preferably using the same kits and laboratory facilities. All unnecessary changes can introduce bias.

Even though the costs of the technology connected to metagenomics’ studies have drastically decreased over the last years, it is still significantly costly. In addition, results are often not immediately available for interpretation, which warrants the opportunity cost of the time as a significant factor in the equation. Nonetheless, obtaining research funding is no easy task and very often researchers are only able to sequence a limited number of samples, so the following question is asked quite often: ‘how many samples do I need to sequence?’. There is no ‘one size fits all’ answer to this elusive question, and it will greatly depend on your research question as we will see in the next page.

In addition, take a moment to reflect on the following question:

‘What types of challenges do you face when designing your studies and how can you overcome them?’