Dr. Hauke Thomsen is a Professor of Biostatistics and Epidemiology at MSB Medical School Berlin, Germany. He joined the Editorial Board of Scientific Reports in 2016 and is a Senior Editorial Board Member. 

His research focuses on molecular genetic epidemiology, genetics, statistics, and bioinformatics to identify the contribution of potential genetic risk factors to the etiology and prevention of diseases within families and across populations. Another field he is working on is large-scale bioinformatic analysis of vector integration sites for gene therapy. He sees bioinformatics as a means to generate biological hypotheses and derive scientific knowledge from computer analysis of complex experimental data. As a statistician related to the medical clinic in Berlin, his work also covers a broad field of analysis in different medical areas. 

***

We recently invited Dr. Thomsen for a Q&A where they answered questions about their research and what it is to work as an Editorial Board Member at Scientific Reports. Some of the questions they answered are below:

• We know that finding reviewers is one of the hardest parts of an editorial role. Do you have any tricks on finding reviewers? 

To avoid long waiting periods, I prefer to invite reviewers with institutional or professional email addresses rather than personal webmail accounts. My selection of candidates is mostly based on choosing experienced scientists: the H‑index should be high, and the number of publications in the last five years should be within a certain range. In my experience, researchers with more publications tend to have a better understanding of the review process. 

• How important is reproducibility in research? As an Editor, how do you help authors report reproducible results? 

Without reproducibility, science becomes opinion; with it, science becomes reliable knowledge. Thus, I always challenge the authors to confirm their results on independent samples if this has not already been done. This helps determine whether findings are robust or instead due to chance, bias, or analytical error. From my own experience in molecular genetic epidemiology, I have faced ‘replication crises’, showing many published findings could not be reproduced; reproducibility reduces this risk. In many cases, i.e. GWAS, uploading data to public repositories, describing methods clearly, and sharing code enhances public and scientific trust. 

• What are the key things journals should do to ensure scientific rigor?   

Prioritizing methodological soundness over novelty, transparency over opacity, and long‑term reliability over short‑term impact metrics. This means that the journal should enforce transparent reporting and strengthen statistical and methodological review. One important requirement is to promote the sharing of data and code. Especially, young scientists should be encouraged to prepare their working hypotheses and present their analysis plans. It is also important to value well‑conducted null results and avoid overemphasis on novelty at the expense of rigour. 

• What would you like to share with your fellow researchers on publishing in an inclusive journal?     

To maintain quality standards, we should be strong enough to also reject manuscripts that have not improved during revision. Personally, I find that if a manuscript cannot be made acceptable following multiple rounds of review, it becomes less likely that an Accept decision will be reached. If comments are not addressed in the next version, the manuscript may have to be rejected. It is also tiring for the editorial board and reviewers to process the manuscript over and over again without seeing any major improvement. 

• What are the biggest challenges that you see for the future of research and research dissemination? 

One of the biggest challenges relates to AI in research and writing. This includes authorship standards, bias amplification, plagiarism concerns, and distinguishing human‑ from AI‑generated scholarship. AI will for sure reshape data analysis, drafting and literature analysis, but we must ensure it does not do all the work including generating the data.  

• Leaning on your expertise in biostatistics and genetic epidemiology, could you share how to approach and assess this type of work? In particular, we would like to know your perspective on the strengths and limitations of the methodology in relation to the specific research question.   

In general, a method is ‘strong’ when its assumptions align with the scientific aim, data structure and causal framework – and limited when they do not. This means that you should clarify the research question first, then apply the appropriate methods. Many times, researchers apply methods or models (e.g., regression models), that seem to result in strong associations but are in fact weak for causal claims. The same holds for GWAS as they may be strong to discover variant associations but lack to explain the biology or the disease architecture. The strength of a methodology lies not in its sophistication but in its fit to the research question and its assumptions being defensible. In biostatistics and genetic epidemiology, methodological rigour means constantly asking: Does this method truly answer my question – or merely produce statistically significant results? 

• Could you also share examples of Mendelian randomization studies that have been carried out well, either from your own work, that of collaborators or papers you have reviewed? What key lessons might fellow editors take from these examples when evaluating MR studies? 

 Overall, I have noticed an increase of manuscripts on MR related topics, and I am somehow critical about the potential of these analyses and the consecutive results. With due respect, MR approaches have some potential to contribute to an improved understanding of the importance of environmental factors in common diseases by reducing the effects of confounding and bias. But there are quite some limitations. MR still has some limitations such as the failure to establish reliable genotype-intermediate phenotype or genotype-disease associations due to confounding of these associations. It is highly influenced by pleiotropy and multi-function of genes and suffers from the lack of suitable polymorphisms for studying modifiable exposures of interest. It would be worth replacing many MR studies through the application of the growing knowledge provided by the study of functional genomics. One can always see that authors try to find a way around in case of limited study material. In many cases they are not willing or not able to collaborate with other groups in order to perform i.e. more reliable meta-analyses. And therefore, their interpretation is mostly vague.

Homepage: https://www.medicalschool-berlin.de/team-fakultaet-medizin/prof-dr-sc-agr-habil-hauke-thomsen/  

ORCiD: https://orcid.org/0000-0001-5951-3116