Q. Could you introduce yourself and your research ? 

I'm Martin Steinegger. I'm an associate professor at Seoul National University in the School of Biological Sciences. And our group has this mission to accelerate biological sciences through fast and easy-to-use software. We are a completely dry lab, so we don't have any wet lab experiments. We're building software and that software is there to analyze the protein universe. So really understanding the protein universe in its entirety. How many different protein functions do we have? How many different protein structures exist out there? This is why we build these tools to tackle these really large-scale questions. 

Q. First, Congratulations on being selected for the ‘Leader Research’ grant by the Ministry of Science and ICT. Could you tell us about this project and your main goals for it? 

So I'm very, very honored to get that award. So it really shows that I think the research that the lab has done, and my students and postdocs and collaborators is really valued here. And I think that I'm very, very thankful. 

The goal of the grant is to build what we describe as an in silico immune system on a planetary scale: a computational framework that mines the enormous amount of sequencing data being generated worldwide and identifies viruses or other agents that may pose a risk to humans. Emerging pathogens and new outbreaks continue to cause loss of life, major economic costs, and broad disruption to society. 

We do not expect to predict the next pandemic with certainty. Instead, we want to identify and characterize potentially dangerous organisms as early as possible, model how they may interact with humans, and prioritize the candidates that warrant closer experimental or epidemiological investigation. 

[Photographed by BRIC] 

Q. A big part of your research focuses on protein structures, and you recently published a paper on Folddisco. Could you introduce Folddisco to us? 

First of all, protein structures are very conserved in nature. They are fundamental building units that you see over and over again in nature that make up the functional repertoire. So sometimes a structures are associated with specific function. And so, we can mine them now. We can understand them. 

Why can we do that? Because now with, like, these highly accurate structure prediction methods, we can predict these structures on a large scale. And there's a resource called the AlphaFold database. And there are over 214 million predicted structures in them. So these structures are not just 3D objects, right? These are really molecular functions. Understanding, like, how biology works. And what does Folddisco do? 

Folddisco finds small 3D shapes that seem to be recurring in these structures. And it does this very fast. So it can take a small catalytic site. For example, enzymes often bind small molecules using around three residues. These residues are arranged in a certain orientation to each other, and when they are in this orientation, they do this function. This means that while structures overall are bigger blocks, you have these really small substructures- and these substructures are critical for understanding functions. 

Folddisco makes it practical to search for such motifs across the entire AlphaFold database. Searches that were previously computationally prohibitive—or could take months at this scale—can now be completed in minutes or seconds, depending on the query. 

So you can actually understand these local functional units. In the end, structures are not just big block, they are consist of really small parts that really make up what proteins can do. 

Q. Since AlphaFold, protein structure data has grown exponentially. You're very interested in ultra-fast searching. And improving efficiency. What was the background of this research and what is the biggest change this tool will bring to academia?

So why do I find these fast searches so interesting? AlphaFold predicted 214 million structures, but what now? What does that really mean? You need a tool that can take all of these 214 million structures and extract biology from them. That's why we built our tools- so we can actually organize this data. 

And why structures? Because structures are very very conserved. Over billions of years, the structures have been retained within proteins that share similar functions. Now it's really cool. We can actually connect the protein universe through structural space. We can group all these structures together and compare them. 

As a result, we can now see structural similarities between humans and bacteria that were never visible before. For example, we have an innate immune system. Unlike the adaptive immune systems, which learns from infections, the innate immune system immediately recognizes a virus and triggers responses like cell death. We don't know how ancient this immune system really was. Using structure comparison, we identified structurally related protein modules in bacteria that could help us understand the evolutionary origins of components of the human innate immune system. These relationships were essentially undetectable using sequence alignment alone. 

With these powerful tools and massive structural databases, I think we will uncover many more fascinating links that help us understand how our own immune system was shaped. 

Q. Beyond MMseqs2 and ColabFold, you have continuously released important tools like Linclust for ultra-fast clustering and Foldseek for structural alignment. Is there another tool or project you want BRIC readers to pay special attention to? 

We work currently on RNA, and we just shared a preprint in the RNA field, which I'm pretty excited about. Protein structures prediction for single chains is pretty mature now - it works really well. RNA also has three-dimensional shapes that drive enzymatic and cellular functions, but we still lack similarly scalable tools for searching and organizing RNA structures and relating them to function. Our goal is to provide the RNA community with search and analysis capabilities comparable to those now available for proteins. I would not call this the near future, but our lab is developing tools that move in this direction, and I am very excited about them. 

We also work currently on the idea of interfaces. Biology does not operate through isolated molecules. Proteins interact with other proteins, small ligands, RNA, or DNA, and these interactions form interfaces—the regions where two biological units come into contact. 

And I find it very interesting to understand how many of these interfaces exist. Is there a language to interfaces? Have we already figured out all of the possible ways how proteins might interact with each other? So we're extending our tools to answer these questions about how molecules interact with each other, and how much do we understand already and how much is still missing, I guess. Exciting too.

Q. Your collaboration with Google also drew a lot of attention. What was the biggest insight you got from that collaboration? 

One of the biggest insights was how differently academia and industry are organized, and how complementary they can be. Google DeepMind brings together outstanding researchers and engineers with substantial computational infrastructure. Seeing how they approach large scientific projects has exposed me to a different research environment. 

We currently have an ongoing collaboration with Google DeepMind, NVIDIA, and the EMBL-EBI. This kind of collaboration can allow academics to pursue questions at a scale that would otherwise be difficult. Academic groups contribute scientific questions, specialized knowledge, and the freedom to explore unconventional ideas, while industry partners can contribute engineering depth, computational resources, and experience building systems at scale. 

I therefore see strong potential for collaboration between the two sides, provided that the scientific goals and expectations are clear. We continue to have active collaborations and work on topics together. 

Q. Do you have any plans to collaborate with Min-kyung Baek, who is best known as the first author of the RoseTTAFold paper? 

We do not currently have a defined joint project, but I would certainly be open to collaborating with her. People often ask whether we compete because both of us work on structural biology, but I see our research programs as highly complementary.  Fundamentally, my lab is interested in large-scale searches and organizing the protein space, whereas she is deeply focused on prediction methods for antibodies, antibody interactions, and really pushing protein design forward. So I don't think we are in competition at all. I think we are on very good terms. 

And here's a fun fact; when I first came to Korea for an internship a long time ago- maybe back in 2012, over 14 years ago, Min-kyung was already in Professor Chaok Seok’s Lab. We actually overlapped back then, and we spent a lot of time together even before either of us became faculty members. I have known and respected her for a long time. She is an exceptionally smart and talented scientist. 

Q. Could you share the Eureka moment of your entire research career? 

I don’t think there is a single Eureka moment. I think every project has its own. Whenever you start a new project you need to solve a new challenge. So, these moments happen to me kind of frequently. Usually when you're very relaxed - taking a shower or in an environment where you feel very comfortable- out of nowhere, you think about your project and you come up with this idea. That idea becomes the missing piece that moves the research forward. And I think they happened really often, I think in some way. So it's not just one moment. 

One major moment was realizing how powerfully a three-dimensional protein structure can be represented as a one-dimensional sequence for computational comparison. The broader idea of structural alphabets is not new, but in Foldseek we developed a representation that captures tertiary residue interactions and combined it with fast sequence-search techniques. This made protein structure comparison dramatically faster and allowed very large structural databases to become searchable. Together with Johannes Söding, my former PhD supervisor, and the wider Foldseek team, we developed the method around this principle. 

Q. So as a teenager, you went to vocational high school and work until you were 25. You wanted to go to university earlier. I heard. But things didn't work out as planned. How did you feel back then? And how did you overcome that frustration? 

My academic way is pretty zigzag, not very straight, and for the majority of time, I did not think I would become a researcher. In Germany, we have an educational system composed of three tracks: lowest, middle, and highest. Being in the highest track is called Gymnasium, which allows you to go to university. I initially attended the school track that was least directly oriented toward university, so at the time it seemed clear that higher education would not be part of my path. Coming from a non-academic family, I did not grow up seeing university as an obvious or necessary destination. However, I enjoyed working on computers since I was 10 years old, trying to get games running and understanding operating systems. 

When I finished my initial education and wanted to do something with computers, the German system offered no options—they told me the only thing I could do was sell computers in a store. But I found my way through Austria, different schooling, and eventually worked at a consulting company. Around that time, the legislation changed, and I could actually enter university with the degrees I had collected along the way. So I thought, ‘'Why don't I give it a try?’‘ 

However, when you walk such an indirect path, you often feel like an imposter—like you don't belong. Going to university, I was older and had work experience, but without the traditional Gymnasium background, you feel like an outsider in some ways. It was not easy to navigate this unconventional path, and the feeling of being an outsider did not disappear immediately. Over time, however, I learned that the different experience I brought with me was also a strength. 

[Photographed by BRIC] 

Q. You lived and worked in Germany, the U.S., Spain before coming to Seoul. What made you choose Korea? 

Actually, I have lived in multiple places and I also have been to many countries. In general, I like experiencing different countries. During my PhD- actually when I was in America- I made friends from Korea, which got me interested in the country. That’s how I got in touch with Professor Chaok Seok, and she gave me the opportunity to be a visiting scientist in her research group. So I stayed here, which was when I overlapped with Minkyung. Initially, I wanted to stay for one year, but I ended up staying for three years.

Even after leaving, the connection persisted. At one point, I had the chance to get a faculty position at Seoul National University. Since I had lived in Seoul before and was already familiar with the university environment, I thought, okay, I can give it a try. I could try to become a professor here. Looking back in hindsight, I think it was a very good decision.

Q. You have experienced research environments in many different countries. What do you think are the strengths and weaknesses of the research environment in Korea? 

Actually, Korea, as a country, has very high per capita spending on R&D. So there is money and resources in the system. As a researcher, when you come here, that means I can start a group, build a large team, and actually undertake ambitious projects, which is great. Also, being at SNU gives you access to really good talent- You get really smart, passionate and hardworking students. 

A weakness is that many administrative processes and important discussions still take place primarily in Korean, which can make it difficult for international faculty to participate fully. This is not a criticism of individual colleagues, my department has been very supportive, but it is a structural issue universities will need to address if they want to internationalize further.

Q. As a foreign scientist, you have settled down well in Korea academic community. So what do you think is the secret behind your successful integration? 

I think it was partly a matter of timing and partly the visibility of our field. AI and computational biology were attracting growing interest, so many researchers were actively looking for opportunities to collaborate with groups like ours. 

I have spoken with other international faculty members who found it much harder to build local collaborations or become integrated into the academic community. My experience was different, partly because I was working in a highly visible field and partly because I had already lived in Korea and had a small network from which to start. 

For someone arriving in Korea without an existing network, family connections, or Korean-language skills, the transition can nevertheless be very difficult. In my experience, universities do not yet provide consistent institutional support for international faculty across all aspects of academic and administrative life. 

At the same time, my own department, Biological Sciences, made a genuine effort to help me. Support appears to vary considerably between institutions and departments. 

Ultimately, I was fortunate to enter a highly visible field at the right time, and the work of my students, postdocs, and collaborators created many natural opportunities for engagement. 

Q. Korea is actively trying to attract top international researchers, making talent acquisition a top priority. Based on your experience, what kind of support or policy do you think is most needed to help overseas scholars settle in and thrive here? 

The single most important change would be to ensure that essential university communication, administration, and faculty governance are accessible in English as well as Korean. An international faculty member should be able to participate fully in the academic system from the time they arrive. 

It is unrealistic to assume that newly arrived faculty can quickly acquire the level of Korean needed for complex administrative and governance discussions. Learning Korean is valuable and should be encouraged, but functional inclusion cannot depend on achieving professional fluency first. Reducing these language barriers is therefore critical if universities are serious about internationalization. 

Q. Some students reading this interview might want to join your lab. What kind of students are you looking for? And what skill or quality are most required? 

We are a technical Lab. So I would ask certain technical questions during interviews. Applicants should have some understanding of how programming works and how computers work. And since we’re also a biological lab, I would be also interested in biological understanding. 

However, I have also hired people whose technical background wasn’t as strong initially, but the person convinced me through the compelling ideas they put forward. If you’re truly self-motivated and can show me that drive and genuine interest, I find that to be an extremely attractive quality.

 I want to see that you have ideas you are genuinely excited about, and that you understand why our group is the right fit for you. Rather than simply chasing the reputation or success of a lab, you should chase the lab’s core scientific idea-figure out why this research excites you and why you truly want to do it. 

Martin Steinegger Lab.