Michael Stadler has been leading the FMI Computational Biology technology platform since 2008. In an interview, he explains how the platform is set up and how it supports the FMI research groups, he shares what he likes most about his job, and he talks about the future of computational biology.
Since when does Computational Biology at the FMI exist and how is it set up?
Computational Biology was created as a technology platform soon after Susan Gasser joined as Director, in 2004, and was initially focused on helping people to find resources on the Internet. I joined in 2006 and started leading the platform in 2008, at a time when the experimental researchers were transitioning from studies looking at a few genes to studies looking at the whole genome, and when the need for data analysis became strong. That’s when the collaboration with the research groups really started. Since then, the need for data analysis - and therefore the size of our team - has steadily increased. Today there are five of us in the core team and another ten or so embedded in the experimental groups. Because of the interdisciplinary nature of our work, our team is very heterogeneous, consisting of mathematicians, physicists, computer scientists and biologists. We have our specific strengths, but we all share the same interest for "biological questions".
Can you briefly describe the main activities of your team?
I would say that they can be split in three:
- We provide an infrastructure, including computers and software, that everybody at the institute can use,
- We teach - be it lectures at universities or workshops at the FMI or externally, and also often on a one-on-one basis.
- And most importantly, we engage in scientific collaborations with our customers, e.g. any experimental group who generates a lot of data and would like to process and interpret that data. We don’t systematically perform all the analysis for the biologists: People who generate data should also be knowledgeable about data analysis. That means that the experimental scientists do part of the analysis themselves and only get in touch with us for bigger and more challenging projects.
Which research groups do you work with the most? Could you name two particularly interesting recent collaborations?
Obviously the research groups who generate the most data! We have ongoing collaborations with 16 (out of 23) research groups. Some groups such as the Peters and the Rijli group even have two people dedicated to data analysis in their teams. Contrary to the past, we now also increasingly work with the Neurobiology groups as they started doing more -omics experiments and generating more data, e.g. from neuronal activity or behavioral experiments with mice.
In 2019 we had some very interesting collaborations related to single-cell RNA sequencing - a new type of data where we are still learning how to best analyze it. An example would be the collaboration with the Liberali group. Another increasing and challenging type of project was the analysis of data generated by "multi -omics studies", where data from many different types of experiments need to be integrated, such as ChIP-seq, RNA-seq, ATAC-seq and HiC-seq, as illustrated by a collaboration with the Schübeler group.
Today most big research institutes and pharma companies have a computational biology group. To what extent would you say that our platform at the FMI is different from others?
I think there are a couple of aspects that make our platform a unique asset for the FMI. First, we established our platform in the early 2000s - quite early compared to others - and gathered significant experience. Then, a great advantage is that our work involves very little bureaucracy, e.g. we don’t have to charge the research group individually for the time we spent working for them. But what is most important in my eyes, is the fact that we have a great team. In computational biology, the demand is much higher than the number of skilled people on the job market - that’s because it’s a relatively new field and the universities haven’t really adapted their curricula yet, and the need has grown exponentially. Therefore, it’s very difficult to recruit the right people, especially in Switzerland, and Basel in particular, where we compete not only with the university and other academic institutes, but also with big pharma. Despite that competition, we managed to attract top talents to the FMI. I’m very proud of my team! Everybody brings along the two key requirements for our job: a strong passion for the scientific questions shared with the customers, and a flexible mentality that makes us happy to work on biological questions of others.
Let me emphasize at this point that we may compete with other institutes from a recruitment perspective, but we also collaborate a lot with them. In an area as interdisciplinary and fast-paced as ours, it’s essential to share experience and best practices with others. For example, my group is a member of the Swiss Institute of Bioinformatics.
What do you like best about your job?
My job is never boring! It’s very much technology-driven; the experimental scientists come up with new types of experiments all the time, generating new types of data, requiring new types of analysis. We always have to learn, adapt, develop new methods.
This is nice, but also challenging since we constantly have to ensure that we are up to date. We also have to continuously optimize our infrastructure, in close collaboration with IT. For example, we recently purchased GPUs, graphic cards which are usually used to play computer games but which are becoming important for our work, in particular in the area of machine learning.
What I also like about my job is the breadth of the science we are exposed to: we contribute to answering a lot of different scientific questions! And then, there is an aspect of my job, which I underestimated when I first started, but which I find fascinating: it’s the importance of the human interaction. It’s not just the science that makes a project successful, but also efficient communication, dedication, endurance… In my team we are confronted with all these aspects as well when dealing with our customers.
How do you see the shortand long-term future of computational biology?
This field will continue to be technology-driven and I believe that in the next 5-10 years, we’ll continue to work similarly as we have in the past 10 years and constantly adapt. For example, new spatial transcriptomics technologies are coming up, which allow to measure gene expression localized in the tissue. This combines transcriptomic data with imaging data, and represents a whole new challenge. Nobody at the FMI is using those technologies yet, but I have no doubt that this will come.
It’s harder to make a prognosis on how computational biology will look in a more distant future. It could be that the major methods to analyze data will have matured and there won’t be the same need any more to develop new methods. In addition, the experimental scientists will learn more about computational biology in their studies. It’s likely that a part of computational biology will be absorbed by experimental biology in the future.