Combining Medicine and AI  [28.05.19]

Actually, she's a nutritionist. Prof. Dr. Melina Claußnitzer uses computer-based models to investigate the genetic causes of metabolic diseases. She is now head of the Department of Nutrition Science in Hohenheim.

 

At the same time, she also has a second foothold at Harvard Medical School and the Broad Institute of MIT and Harvard in Boston - a balancing act that also benefits her Hohenheim protégés.

 

Ms. Claußnitzer, you have actually been in Hohenheim since October 2017...

... officially, yes, but then our second child was born, so I only started here in April 2018. That was also a while ago, but it was like the date was jinxed: Every time something came up for one of us and we had to postpone it.

 

...but now it finally worked out, that's great. You are the successor of Prof. Biesalski. What will change with the department's focus?

First of all the name has changed - from "Biological Chemistry and Nutrition Science" to "Nutrition Science". The focus will change quite a bit. Mr. Biesalski is an expert in the field of nutrients and vitamins. I also studied nutritional sciences, but now I have a new research focus which is still very rare in Germany.

I investigate metabolic diseases using methods from the field of computer science: Machine Learning, Artificial Intelligence, and Big Data. The nutrition sciences have changed considerably, the approaches are more personalized and one-size-fits-all is no longer valid today, because it simply does not fit all people.

You studied nutritional sciences here in Hohenheim?

Yes, my Diplom thesis, supervised by Mr. Graeve and Ms. Daniel from the Technical University of Munich, already had a strong mathematical background since it was about modelling transport processes that are relevant for metabolic diseases.

Where were you after that?

Then I went to the TU Munich for my doctorate, to the chair of nutrition medicine under Prof. Hans Hauner. I cooperated closely with the Helmholtz Centre as part of the Virtual Diabetes Institute. The aim was to decipher genetic signals that are crucial for type 2 diabetes.

Since the human genome has been deciphered, many diseases can be better understood, but the topic is very complex. Together with other scientists, we have discovered that the signals do not lie in the genes, but in regulatory elements. They were previously regarded as so-called junk DNA, i.e. gene segments without any function. However, about 93 percent of all genetic signals result from this.

You were also talking about computer science - where did that come into play?

During my doctoral thesis on 'Functional characterization of type 2 diabetes associated genetic variants', it quickly became clear that the large amount of data could only be interpreted with the development of new computer-based methods. The idea was to use bioinformatic methods to identify causal genetic variants in junk DNA. When I presented this bioinformatics method at the Gordon Research Conference in the USA, I was offered a postdoctoral position at Harvard Medical School in Boston. The work dealt with age-related diseases, including type 2 diabetes, obesity, and Alzheimer's disease.

At a computer science conference, I got into a conversation with Eric Lander. He was involved in the worldwide human genome project and made a major contribution to deciphering the human genome. He said our methods and approaches were important, and since then I have divided my postdoc: Half of my time I continued at the Institute for Aging Research at Harvard Medical School, half I worked at the Computer Science and Artificial Intelligence Lab at MIT. That was sometimes stressful to do justice to both, but it was also the most important step in my scientific life. This later gave me the opportunity to move my research group to the world famous Biotech Institute, the Broad Institute of MIT and Harvard, where I have been an Institute Member ever since.

Why was that step so important?

These two sides are rarely united in one person, they often do not understand each other at all. But this opens up completely new possibilities: The models we work with are agnostic. So we're not assuming a hypothesis. In this way you often end up at a point you didn't expect.

For example, we have used mathematical models to investigate which genetic signals are relevant for people who are overweight. Until now, it has always been assumed that the crux of the matter lies in the brain, that people do not feel full. However, our work has shown that it has nothing to do with the brain, but is due to fat tissue.

Humans suffer a cold shock at birth and have to produce heat suddenly. Newborns, on the other hand, have so-called brown fatty tissue that quickly releases heat.

There are different fat cells?

Exactly. White fat cells are responsible for storing energy in the form of fat. And genetic markers have enabled us to discover a new, previously unknown type of fat cell: beige fat cells, which, unlike brown fat cells, are also found in large quantities in adults. The white fat cells are precursors and can develop into beige fat cells. The beige fat cells resemble the brown ones in some respects, and they can also burn energy relatively quickly. This is controlled, among other things, by a certain hormone, and we hope that this can be an approach for the treatment of obesity.

That was indeed an unexpected result...

... yes, and it shows how important an agnostic approach is. We have joined forces to form the Common Disease Consortium (CDC) and now want to carry out a major project worldwide. In 5 years we would like to clarify the genetic basis of the 5 most important metabolic diseases such as Crohn's disease or type 2 diabetes and develop treatment options for them.

And you are now establishing this research in Hohenheim?

Yes, although that's not easy. Here in Germany it is difficult to recruit computer scientists and experimentalists who are familiar with Big Data. Qualifications in artificial intelligence in medicine are rare. Things are going slowly, but it's working. Therefore, and also because the conversion of the laboratory rooms here in Hohenheim has not yet been completed, I am glad that I still have a foot in Harvard at the moment and a laboratory there as well.

You run two labs in parallel on two continents?

Yes, the spatial distance is no longer so important today. By the way, the team management app Slack is very helpful for the fusion of the laboratories and I can only recommend it. It works very well between Boston and here. And then the two locations are definitely also an incentive.

On this occasion another aspect which is very important to me: We are very inclusive, open to all ethnic groups, sexual orientations etc. My laboratory has always been colorfully mixed in every respect - and that has always been an enormous enrichment!

Ms. Claußnitzer, imagine that you could have unlimited resources and opportunities - which project would you start?

I would simply speed up the project under way with the Common Disease Consortium. We are investigating 60,000 genetic regions for causal relationships with diseases and have initiated 33 projects. Research is very expensive, which could of course be accelerated with the appropriate funds. It would be good if we could decipher all genetic signals for certain diseases as quickly as possible and use them for the prevention and treatment of diseases that are threatening for all of us. Here I consider the research of genetic signals of all ethnic origins as essential. At present, genomic studies are carried out almost exclusively on Europeans. That must change.

With our Variants-to-function (V2F) initiative, we want to find out which genetic variants are associated with diseases and which therapies can be derived from them. I would like to integrate this initiative here in Hohenheim, where discussions and initial projects with colleagues are already taking place.

Let us come to teaching. What can we learn from you?

In teaching, not so much has changed for the time being, we continue to teach the basics of nutrition science as well as nutrition-dependent diseases like Mr. Biesalski did. But we are now moving more in the direction of personalized medicine, and we will continue to expand this approach. The students' interest is definitely there.

Can students participate in your research projects?

Yes, I already have quite a few requests for Bachelor's and Master's theses. If possible, we then involve the Master's students in ongoing research projects. We publish relatively little, but in very renowned journals - and Master's students are part of it.

In the summer we had a workshop with Norwegian colleagues, the Broad Institute, and Stanford University. For 5 days, the focus was on the core projects, and some students were also invited. Some people caught fire!

I also have Humboldt reloaded in mind, which I will have to take a closer look at in the near future.

What is the hallmark of good teaching?

Generate enthusiasm for progress and new developments. I'm excited about the research myself. These new aspects must be introduced into teaching, that's what should dominate the slides - not the textbook knowledge. Primary literature is also important.

Teaching must be alive. The type of exams in Bachelor's and Master's are not exactly conducive to finding one's vocation. But I hope that a handful of students will remain enthusiastic, and that we can at least take the others with us and create good foundations for their professional careers.

Keyword profession: Where can your students work later - apart from research?

There's a wide range of applications. In industry, for example, in pharmaceutical companies. The industrial sector pays quite well, they often snatch the best people from us...

But you can also go into science journalism, for example, or into nutrition counselling in the traditional way. That's what I wanted to do before I got caught up in research. The personal factors that we now have on the screen are also very interesting for practice.

How has Hohenheim changed since your student days? 

Studying is stricter now, with more time pressure. That's why I miss the Diplom programs a bit. But the level of the students is still excellent, the people are very ambitious.

I immediately felt at home again in Hohenheim, I simply love it. I have gotten to know many universities, but the Hohenheim campus has the most beautiful charisma of all, you can' t find it anywhere else. I only miss Wittwer and the small grocery store that was there in the row of shops - although Denkbar is not bad either.

By the way, the handling of appointments is also fantastic, the recruitment of young professors - everything is much more progressive than back then. There is also a generation change, and the young generation has very few pretensions.

How do you spend your free time, Ms. Claussnitzer?

I have two small children, and I'm a passionate mother. I prefer to take the whole family with me to conferences - so far the children are still doing well. As an Allgäuer I also love the mountains and hiking. Paragliding and climbing used to be on the program, but that's no longer possible with the children. I love nature and also need it to be innovative - I have developed many ideas and methods there.

 

Thank you very much for the exciting interview!

 

Interview: Dorothea Elsners, Translation: Neudorfer