Posted By: The American Journal of Human Genetics, AJHG
Each month, the editors of The American Journal of Human Genetics interview an author of a recently published paper. This month, we check in with Adriaan (LinkedIn; BlueSky) to discuss his recent paper, “Mendelian randomization linking metabolites with enzymes reveals pathway regulation and therapeutic avenues.”
AJHG: What motivated you to start working on this project?
AV: There’s a statistical genetic technique called Mendelian randomization, which can very easily identify if there is a causal link between two traits, as long as there are genome-wide association study (GWAS) summary statistics available for both of them. That’s a pretty big deal, as identifying causality between two traits in humans usually requires expensive and time-consuming randomized control trials. Because Mendelian randomization is easy to perform, many low-effort studies have applied the technique in questionable ways, which unfortunately reduces confidence in the technique. This is especially worrisome as there is a knowledge gap: we simply do not know how often Mendelian randomization comes to the right conclusion. This makes it even more difficult to assess the true merit of these papers.
We were wondering if we could address this knowledge gap by applying Mendelian randomization to causal relationships known in humans, trying to figure out when Mendelian randomization is right and when it is wrong. To identify a ground-truth dataset, we turned to a very well-known human biology: metabolic pathways. Recent large studies have determined the GWAS summary statistics of both metabolite concentrations and protein abundances in humans. By combining ground-truth metabolic maps, protein studies, and metabolite studies, it becomes possible to systematically identify when Mendelian randomization works and when it doesn’t.
AJHG: What about the paper/project most excites you?
AV: What excites me most about this paper is that, when applied properly, there seems to be a lot of merit in Mendelian randomization. We can match human metabolism in up to 43% of positive findings. We think this is actually a lot, as not all biological links have been rigorously studied, suggesting that some links lacking prior evidence may actually represent previously unstudied biology. On top of this, these links are not only interesting for understanding basic human biology, but they are also very important for drug development. We show it is possible to identify meaningful links between drug target proteins and metabolites. These causal links provide evidence for the intended therapeutic effect of perturbing the protein, and potentially even what kind of side effects can be expected from targeting the protein. Finally, these results also allow us to understand which ground-truth links are systematically missed by Mendelian randomization methods, which may help us further improve these statistical methods.
AJHG: Thinking about the bigger picture, what implications do you see from this work for the larger human genetics community?
AV: I think this work will provide researchers a more clear picture on how to assess Mendelian randomization evidence. When performed correctly, Mendelian randomization identifies robust causality between two human traits, which aids drug development, biomarker identification, and basic human biology. This is of course not to say that Mendelian randomization is the only tool necessary to determine if two traits are related. It can, however, be a very good tool in the toolbox of statistical genetics. A notion we hope that the genetics community will embrace.
AJHG: What advice do you have for trainees/young scientists?
AV: My advice is first and foremost to try and identify a research group that fosters your development. Unfortunately, I’ve too often seen a young researcher join a group because they are passionate about the subject matter, but the group fails to provide a place for the researcher to thrive. If a research group is passionate about you as a researcher, this will make the research so much more fun, and very often, it becomes possible to point your research to the subjects that interest you.
I’ve had the good fortune to be supervised by many excellent supervisors who have helped me thrive in research. These include: Prof. Johannes, Prof. Janssen, Prof. Franke, Prof. Visscher, Prof. Sanna, Prof. Wijmenga, and my current supervisor, Prof. Zoltán Kutalik, PhD. I’m extremely thankful to them for letting me do this research.
AJHG: And for fun, tell us something about your life outside of the lab.
AV: Outside of the lab, I enjoy running middle to long distances. Currently, I’m training for my second marathon. Creatively, I design and construct my own lamps, which started as a hobby during the pandemic, allowing me to make something physical. If you’re interested, feel free to visit my website.
Adriaan van der Graaf, PhD, is a postdoctoral statistical geneticist at the Statistical Genetics Group of the University of Lausanne.