Conversations with Dr. Hyungwon Choi

Unlike 20 to 30 years ago, endocrinologists, neurologists, cancer biologists, and the like work with high-dimensional molecular data all the time. Measuring human gene products alone gives you so much information – how do you deal with that information? They need somebody who can crunch the numbers for them so that they can pinpoint what is the most important information in a particular patient. I help them to see that information buried in a very large amount of data.

This type of data is typically inaccessible to most people because you need to work with the people who can generate high-tech molecular measurements. When it comes to obtaining such information and using them in practice, there are, in the U.S. for example, companies to which you can send your biological samples, like sputum or blood samples, and they will analyze your DNA. For example, the company 23andme is able to conduct genetic testing and report back your current and future risks for many common diseases. They calculate the probability of you getting a disease based on your DNA profile, by cross referencing it to a large database on the U.S. population. Based on your genetic profile, they can tell you if you are likely to contract certain diseases by certain age. This – building a model for predicting health risks – is part of what I do.

Apart from demographic data, I also deal with data at the molecular level. For example, if you get a panel of biochemical tests of your blood, you should be able to evaluate your state of health, such as by how much you deviate from a healthy person. But there’s so much irrelevant data you collect that you wouldn’t know what to look at. So my job as an informatician or data scientist is to work with people who generate these very complex data, and to tell you which part of that big data is informative for your investigation. In a way, it is a pattern recognition problem.

Well, I started off as a junior faculty member in NUS, where I taught some medical students and students in public health about molecular epidemiology. I also did some work for discovering biomarkers to determine your predisposition to diabetes or common diseases. Then, I realized that I was more inclined towards pure science. So, I got a part-time job at A*STAR to work with basic scientists, mostly molecular biologists. I had a lot more fun than I ever did when I was working with cancer biologists,. But as time went on, I had to work with the medical doctors in NUS as well. This is a very specialized job though – too narrow to be a career goal when you are young. But I think it is just an example of how you can learn different disciplines of data science. They include Computer Science, Statistics and Mathematics. You can actually apply what you learn to a real problem solving.

It can be pretty hectic working concurrently at both institutions! This is my third year in A*STAR and the scientists are very demanding, but they also understand that I have other obligations. Even so, my lifestyle is probably a little too focussed on work for some people because I wake up usually at around 5:30am and go to bed usually around 1~2:00am. But I enjoy much of the process.

I usually get up and check emails in the morning. When I have time, I walk my dog, then I go to work by about 9. I usually spend my mornings in meetings – most of them with graduate students who do the actual work. I am currently at the stage where I oversee a team of people; I don’t write computer programs. In the afternoon, I try to have time to myself by reading papers, writing some grants, etc. 

When I have lectures, it gets even busier because I have to be in the classroom to teach, and to prepare. Then at 5 or 6pm, on a good day, I go back home, spend some time with my family. After dinner, I sometimes go back to the lab and do my things. At times, for up to a week, I have to travel to conferences. Traveling to the U.S. takes roughly 20 hours. Sometimes I attend a conference for 3 days and then I have to come back for teaching. It is crazy, but I enjoy the process.

When I solve a problem! When I find a panel of protein markers that can actually be used for patient diagnostics, I think that is really rewarding. I have my own share of different kinds of pleasures that I get from my work. For example, when I publish a paper or when I get a grant, it feels great because it means that I can sustain my team for a longer time period and have exciting science projects to do. Having said that, I think that the biggest joy that I get out of this is, of course, the product. When I see that my analysis of a big data leads to seeing 5 or 6 genes that are really critical in explaining the mechanisms of disease onset, and they can immediately be used in a clinic, I think that is the biggest payoff.

Yet, most of the time, it fails! You face setbacks and failures in research. I think being able to deal with failure is the most important thing in any career. A lot of people take it very personally. Even my friends.

Though you would probably get numb to the failures, there are projects that you will value more. And if those fail, of course it feels like a dead end. Should I stop there? I obviously have better options going out to the industry to get paid better for a much smaller amount of work. But I think you should not really care so much about money.

If you are too smart, usually, you do not stay in academia. Surprisingly, being a professor is actually not the smartest decision you can make in your life. You should really love what you are doing because there is no extraordinary financial payoff, your life will be full of failures, and your friends will be making more money, having a much easier and relaxed life, while you are in a lab doing stuff that don’t work.

You have to learn to deal with it. For me, I simply just don’t care anymore. But it takes time and effort to get to that level – it is hard not to care. When I started my job here, applying for research grants was very different. It took me easily 2 or 3 years before I got to the stage where I could very efficiently write a proposal that would actually get funded. So during that time, I was extremely stressed because nobody could help me. It is my own writing. It is not a problem that you can easily fix either. But you learn to make it work, and once I realized how to make it work, it was relieving.

It is not an easy job, I don’t usually recommend being in academia unless you are very determined to stay. And the thing is, a lot of undergraduates I meet think that being a professor is a cool thing – having an easier life. But what they don’t realize is even after securing that position, the stress only gets worse.

I always tell them that you can solve better problems in a much better way outside in the industry. If you get a job in a company, it feels like you are not going to do what I do, but you can actually do way more than what I do. Because they have resources. You can do science outside in companies. I think a lot of the natural science students don’t realize that. They think that everything has to be done in a school environment. But no, actually. Drug companies can do better science.

It is, and it is only getting more competitive.

Also, the kind of projects getting funded is changing rapidly and adapting to such change takes a long time. As researchers, what you learn and when you do your Masters or PhD is very likely to be irrelevant 20 years later. That is the challenge that we all face. We always get slowed down because we don’t see how the real world changes out there. We are actually one of the dumber people out there because we think everything we do is right. We always think of it as a right or wrong problem, but it is not like that in life. I think that is an odd irony. Once you become a professor, you think that you see things better. But you actually see worse. I can definitely tell you. We are so stalled in what we know now that we don’t see what is coming 10 years later.

When I was an undergraduate, cool mathematicians would always tackle the fundamental theoretical problems that you can solve on a piece of paper – nothing computational. As I was studying, things changed in a way that being able to deal with a very large volume of data and doing fancy modelling became more important because many theoretical problems – things that humans can solve – have already been solved. Now, we are living in a time where you can use computers to solve those problems.

This is one such paradigm shift that I experienced. We all have to adapt to new paradigms.

I didn’t know better – I simply had no guidance. My father, a chemical engineer, was in the industry and he tried to discourage me, but I didn’t understand his words then. I was so adamant that I would be doing this.

But I was misinformed in a way. Nonetheless, I don’t regret my decision a lot. I could have easily chosen to go out there, but for some reason I didn’t.

I did use to consider crossing over into something else. When I came over to Singapore, I had 3 choices. One is being an academic in the U.S., one in a drug company in Seattle, and here. Eventually, I came here because of my family. I did have a pretty good offer out of academia and a pharma company. Sometimes I think I could have gone there and probably had an easier life in a way. But I don’t think that being in the industry is necessarily easy. To survive in a very competitive environment, you’ve got to handle a different form of pressure.

When I was younger, I lacked the confidence in my skill sets. I felt that going into academia – following this 10 year process where you go from an assistant professor to an Associate Professor, and guaranteed your research tenure – was the easiest path I could take because of stability.
My father, he had very fierce competition at work all his life. He was a chemical engineer working on industry products and there was always so many competitors out there that will play a game with him in a way. So, he lost his job once when I was young. I think I was 17 then. He then found another job soon, but before then he probably thought that it was almost the end of the world or something. I always saw him drinking. So I didn’t like that. I wanted a stable job and the faculty job was something that I was comfortable with.

Actually, it is a funny story. I went to college as a Spanish major. My mother was a high school French teacher; she always encouraged me to read and write short critiques on novels and things like that. I was inclined to learning languages, so I knew that I could actually be a literature or at least a language major. Anyway, I had very few choices in the Korean school system back then. Because I was in a foreign language high school, I could get all kinds of benefits that would help me get into good colleges by choosing a language major. I actually applied to the French department and I didn’t get it. I got bumped to a Spanish major, which I actually thought was fine. I liked speaking multiple languages.
Then, we had our national service in Korea. My service was 2 and a half years and it was a really good time, actually. There were a lot of times where I wasn’t doing anything, so I had time to reflect on what I wanted to be.

I chose to serve in the middle of the college because I wasn’t sure what to do there. I was learning a lot of history, language, philosophy, but that was definitely not where my passion was. At that time, I knew that I wanted something else. Even before I went into the service, I did some economics and mathematics just for the sake of trying, and of course I did horribly in it. But it was a lot more fun than language. And then, somehow I was hooked on to statistics because it was about data. I loved programming. Writing programs. I learnt my first computer science language back then – C Language – for the first time.

I had a little Apple computer when I was growing up. My father was in that kind of industry so he bought the small Apple computer with the green screen and the keyboard that was so old fashioned. I liked programming, so it occurred to me that this was something that I could do for a long time.

I don’t think I’m a very good statistician, mathematician, or biologist. But I think that I know how to work with those people better, and that is a huge competitive edge. I know people who are way better mathematicians and statisticians than I am, but they usually cannot work with other people. They can definitely not work with biologists because they don’t understand the jargons. It even seems that they tend to refuse to do that on purpose. Perhaps it’s a human tendency to not gel with other disciplines if you become really good at one thing. But my strength is exactly that I can work with many people from varying disciplines, and come up with a product that combines them.

I think I would have tried art. I was into music a lot and if I was better informed, I could have done some sort of engineering plus music together, because I like the acoustics, even the theory of it.

You can be a very good math major and do acoustic music. You go to a rock concert and all those sound combinations that are being mixed actually go through a lot of electrical manipulations and the equipment hooked onto the instruments are very high tech. You have to learn a lot of things. Not only should you have to play, but also, the science behind it is amazing. I could have done that I think. I would have enjoyed playing guitar or piano and come up with a sound that many people can’t.

I went to a concert when I was young. There was a group called Rush. It is an old rock band and when I was 13, and they had a little show of the backstage where you can actually walk through and see their instruments. For the first time, I saw what they call a mixer. It’s a small machine that you put on top of a mixer and it was the coolest thing ever. They were demonstrating that by the stroke of a guitar, you can actually cook up 2 or 3 different sounds from the same chord, depending on how the mixer adjusts the sounds. There is one sound that goes very deep, like in background music, versus solo. It is the same guitar with differing mixing techniques! I didn’t know that. But that was my first revelation when I saw the instrumentation, it was my aha moment. I think I would have loved that. And I wouldn’t be here. But being a scientist is also cool. I like it.

Writing. Being a good writer is very important for scientists. I think it is true for everybody actually. Even when you write a very simple email, I can tell now that I am old and experienced enough, there are people who can express their knowledge and their opinions in more efficient ways than other people. I think those people stand out almost immediately. When I write emails and exchange them with my colleagues, those people who have very clear thoughts as opposed to the ones that have really dispersed way of expressing themselves – I mean I am always confused with these kinds of people. Not that I’m a very good writer, I did get better, but I’m still very average in terms of making people understood. I always see really good writers and they are very successful in what they do. Maybe in research, even more so, because we have to write research grants, and grants are a clear example where you have to sell your ideas to other people. You have to do a sales pitch to convince non-experts, and that’s the hardest bit.
Also, I think computer language – being able to write short scripts, things like that – will be a huge advantage, not only for me, but also for most people in your time. You can write a small app, for instance, and that’s a really powerful thing, because much of your life happens on your electronic gadgets.

One of the things that I think is very cool, is what they call structural biology – structural, 3D imaging of a molecule. Mapping how they fold, change conformation, and such. This goes way back – I’m talking WWI time. People had this technique called X-ray crystallography, where they shoot a laser at a molecule at very specific angles, and you have detectors recording the resonance of diffracted light. You then try to come up with a possible 3D structure and configuration of atoms for the molecule. This was something only the really smart people of the 18^th and 19^th century could do, because only they had access to the machines and the limited data that they had to do deductive reasoning with.

Now we have a ton of data and high-resolution 3D images that can give you clues as to how very long proteins – which are very hard to find the structure for – can be solved by computational algorithms. But you have to be able to write the program, think of the optimisation algorithm, and more. This is really the next generation stuff that not many people can do. It’s challenging enough for a computer science major, but you have to know all this small biochemical details about protein folding, to even come up with the basic idea. And not many people can do it, that’s the thing, and I think this is the area that needs a lot of filling by probably the smartest flock of next-generation scientists.

But, having said that, I think there are other areas that need a lot of workforce. Like, part of what I do for instance, is bringing all this biological knowledge into the numerical data that can be used by medical doctors for example, constructing a large database of medical records that can be queried in different places. That’s like a Google-like problem. Let’s say you walk into a clinic somewhere in different places over the years, and if you can query someone else’s data and look at his or her medical history – what drugs he/she took – and infer what drug you should take, because your physical aspects such as weight and blood profile are very similar to that person. I mean this is like another Google product, and you need a lot of people to do that.

When I was in college, I had no idea, so I didn’t do anything much. It was only after my Masters programme that I wanted to work for something, and I did, in the genetics department. In my 2^nd year of my Masters I started to look for research positions in my university, and there was this perfect job for me – it was a very simple job, helping a geneticist to crunch simple data from a database. That actually landed me a job after a year – after my graduation. I worked in a genome sequencing centre that sequenced the fifth human chromosome for the first time!

If I had known my path, I would have gone for more internships, definitely – even every semester! But, in college, as you can tell, I was quite lost – I did this and that. I thought I would completely fail everything, because I wasn’t that good at anything. I’m serious! Maybe because I was looking at graduate school as my end-goal, and because I had not taken fourth year economics class I was scared to go to graduate school, but it somehow worked.

I think it’s important to be good at the fundamentals when you’re young. When you go to college, don’t choose to an applied subject. Try to become more of a fundamentalist. You know, you learn physics, or mathematics, or core theory of economics, rather than applied economics. For engineering, maybe for example, go learn signal processing, don’t go to data science immediately, because you don’t have a proper foundation. So I think, until you get old enough, just pick one subject and dig deep, and aim to be really good at it at that level. And then if you feel like that’s not your calling, then you can actually go into more applied subjects after that. You could work for Google, or Microsoft, be part of a team. You can actually shine off of the fundamentals you learn. You cannot learn the basics when you are 26 years old. That’s usually not going to end up well, because you’re usually going to end up changing a lot things.

So even if it was going to be very painful, I would have chosen chemistry or physics or even sociology or philosophy. I would actually dig deep, because that’s the only time you get to learn your fundamentals. But you’re not learning the material itself, you’re learning how to learn. Even after college, you’re not an expert in something. But what you learn is how to learn better when you go out of school. So I think, luckily, whether I intended or not, I scratched the surface on many things, but I only learnt to dig deep in my PhD. And even then, I didn’t have very good fundamentals in anything, and that was something that I regretted a lot. If you have an opportunity and good guidance, I think that’s what you should do. Learn something really hard, and try to be a good problem solver within that range, and then you go out there and do your stuff. I’ve rarely seen people who knew what to do from college to their next job. A lot of people get confused as to what they were going to do next.

For instance, I actually have a friend – a physicist by training – who was really smart, and he’s now a banker. And a lot of people do that these days! And he actually got a second PhD because he loves being in school. He went to University of Washington for his physics PhD, and then gave up because he couldn’t solve the problem he was trying to solve. He graduated, but he didn’t manage to solve it, so he applied to another graduate school. He was 34 when he graduated! But he got a decent job, he enjoys what he’s doing – being a risk calculator for a big investment bank. And he loves doing that, because he can see how things are moving from a very different angle. Different from people with a sales background –  because they don’t know the theory, and they don’t have the insight. But this guy, in the same breadth, he’s speaking everything in terms of numbers. He has simulations running in the background, so he can predict what the immediate future will be like, and that’s because he’s so good at the modelling that he learnt.

Yeah, for advice: dig deep, and fully understand when you have the chance. Do that when you’re young, and that will equip you for the future. It’s a hard lesson that I learnt. Actually one of my mentors once told me the same thing when I was young a similar line of thinking, but I never understood – it was too abstract. I guess everyone has to have his own share of experience to get there right, I mean it’s an epiphany at some point. 

And, finally, be yourself!