Night Science

1 | Ellen Rothenberg: inhabiting the data

Itai Yanai & Martin Lercher Season 1 Episode 1

In this episode, your hosts Itai and Martin talk with Ellen Rothenberg, a Distinguished Professor of Biology at Caltech, who always wanted to be Beethoven when she grew up and who feels claustrophobic when doing something that other people are doing. Ellen is one of the leading scientists of our time, and her infectious energy and enthusiasm for science make her an amazing guest. Ellen loves to use metaphors and likes to imagine that she’s a transcription factor in a cell’s nucleus. She stresses how a detailed and explicit knowledge structure is crucial, so that you can recognise an interesting piece of data when it hits you. 

Ellen researches the molecular mechanisms responsible for the decisions made by stem cells as they develop into a type of immune cells. This is a complex process that offers unique insights into the nature of "stem cell-ness". Ellen has won many awards, including the Richard P. Feynman Prize for Excellence in Teaching, and is a Fellow of the American Association for the Advancement of Science and the American Academy of Arts and Sciences.

For more information on Night Science, visit www.night-science.org .

Martin   

Today with us is Ellen Rothenberg and we are very excited to talk to you, Ellen, about your night science. Ellen you received many honors, and just to give a few examples, you're a fellow of the American Academy of Arts and Sciences, you were recently elected to the American Association for the Advancement of Science, and what I personally find extremely impressive is you also won the 2016 Richard P. Feynman prize for excellence in teaching.


Itai  

Absolutely, and I think what is actually even more impressive is Ellen's research, which has made her I think one of the leading scientists of our time. In her research, he studies the mechanisms of vertebrate development, and more specifically, how it is that different classes of T cells develop. On a personal note, I can say that I've met Ellen at many conferences, and the thing that sticks in my mind the most is just her infectious energy and enthusiasm for science. And so, Ellen, it's really such an honor and a pleasure to be talking to today on this podcast.


Ellen 

I'm thrilled to be here. And thank you so much for inviting me,


Martin   

Ellen, just to get started. Do you think there's any aspects of your training as a scientist that influenced your development in terms of becoming a creative scientist?


Ellen  

Well, my training as a scientist sort of ironically began before I was in college, because my parents are both academics. And they always encouraged my interest in science, from a very young age, but at the same time, they infused it with an interest in the arts and literature and history. So that I was brought up sort of immersed in an artistic culture, and I always wanted to be Beethoven when I grew up. So I think that the sense in which the emphasis was placed on creativity really came from long before I started to get formal training as a scientist. I don't know how closely I've succeeded in bringing creativity to science, but I hope I've done some.


Itai  

In what fields were your parents in academia


Ellen  

They were in in social science, they were Franklin D. Roosevelt type economists. But they were very interested. My mother was a superb classical pianist. My father also was interested in classical music and art. And so we had a very strong discourse in the family about art and creativity and what it took to be creative. So going into science, that still was very much part of my set of goals, I think. And I don't think that anything that I learned in Harvard or MIT was able to completely wipe that out.


Martin   

That’s fantastic.


Ellen  

It was a joke. I actually, I had marvelous, marvelous, marvelous teachers at Harvard and MIT. So that was just a joke.


Martin   

Can you distinguish that mode of creativity that you learned as a child or as a teenager, in your work now?


Ellen  

I think there's just a sense of great dissatisfaction with doing something that other people are doing. It makes me feel very claustrophobic, if we're basically following someone else's program. I'm not sure whether it's doing a great service to my students and trainees. But it certainly does keep shoving us against the boundaries.


Martin   

No, and I think it's a great recipe to do groundbreaking research. There's this saying that what a lot of scientists do is they're enlarging a hole that someone else started digging. But if you want to find something really different, you have to start digging a hole somewhere else.


Ellen  

Yeah. Or you can find the fracture planes where there might be some good places to start a new hole. Yes, I think it's a little bit more than that. There is a requirement to be very ambitious in one's thinking in order to see where the current structure of knowledge doesn't fit. And I think in this sense, not in all senses. But in this sense, I kind of agree with T.S. Kuhn on the structure of scientific revolutions, that a lot of times, you do reconfigure the way you understand something, when you realize that the prevailing structure of explanation doesn't fit. And the interesting thing about that is that what really helps is to be very explicit, very clear and very detailed about what the prevailing structure does fit. And that way, when something really doesn't fit, you can see that fact. I think one of the things that I worry about with some younger people who've come into our field at a time when it was so data rich, they actually have a kind of a passive defeatism about that. They don't necessarily feel that they ought to be able to understand everything. And therefore, when something comes in, that actually could be the key to a major reconfiguration, they don't see it because they don't have a well enough articulated sense of what it is that it would violate. Am I making sense?


Itai  

Yeah. And I'm just wondering how we could kind of think about how it should be taught. How would you convey that to someone?


Ellen  

Yeah, so for myself, one of the great educational experiences of my life was a decades-long friendship with a very strong-willed character, Eric Davidson. 


Itai

Yeah, I knew you were gonna say that.


Ellen

Yeah, well, because he was very interested in gene regulation, and how it specifies development. But he was extremely strong-willed, but also very, very ambitious. And he had an extraordinary grasp of a lot of knowledge about embryos. And we used to have great arguments, because he had a very large conceptual structure that he had crystallized over the years, based on information from the way embryos do things in terms of developmental specifications. And so by having someone with such a clear detailed sense of what things should look like, you actually really see when you've got a result that completely does not fit that. And it was fabulous to see the detail of the number of cases in which embryos are doing something one way, and the hematopoetic cells that I was studying, did something different, something that was not expected. So I think that that was a tremendous help, and I think these arguments were very valuable, these inter-field arguments because different biological systems actually optimize different things, and seeing that so clearly was a tremendous, tremendous help.


Itai  

Yeah. And I think in general, the way the human mind works is, we can only make contrasts, something's only defined by how it's different from something else. I can imagine you thinking about development in the immune system, and you had certain ideas, but only when you can see that they really contrast very sharply with what. Eric Davidson is saying then all of a sudden, you say “Oh, actually, that is that is actually very novel and interesting”, because it's not what you would expect.


Ellen  

Yeah. And, you know, it's that difference in the context of overall similarity. And it's also the detail on the picture so that you can detect not just one hole that somebody started digging, as you were saying earlier, but you can detect the whole structure of little cracks, that give you opportunities to go through dose-dependence, flexible timing, stuff like this. This is the heart and soul of hematopoietic  development, but it is anathema to an embryo. So it's been very, very interesting there.


Martin   

Yeah. But it's very interesting that you said that you need some strong-minded, I mean you didn't call it an opponent, right? You call him a friend, right? But somebody who, who has a strong idea of like, you say how things should be so that you can contrast your findings against those ideas.


Ellen 

Yeah, it's a knowledge structure. I mean, you know, I'm not saying that he's wrong about what he worked on. But it was such a detailed knowledge structure with so many pieces to it, that fit together in that system, that you could then see all of the places where your own discovery of something different would have implications. And so that gave you a lot of chances to go off and explore.


Martin   

So, so far, I think you talk mostly about the background against which you develop your work. But when you actually come up with a new idea, with anything creative, do you think there's a specific method that you use to do that?


Ellen  

I wouldn't put it that way. Martin, I think most of it is inhabiting the data. So again, if you are really trying to get a full understanding of a mechanism, then every piece of data that you get either fits or it doesn't fit. And that can be, you know in the old days, we used to do experiments that were really hypothesis-based, and where every assessment of the samples that we made was based on a very focused hypothesis. Now we do a very large fraction of our work by sort of genomics methods, which gives you this huge harvest of data, which wasn't explicitly part of your initial hypothesis, and it's there freely for you to play with. I think I drive the people in my lab crazy looking at genome browsers and looking at gene lists, because I'm always trying to cluster them and put them together into pathways that I think I already know about. And I start jumping up and down all the time, when I can find something that looks as though it's showing us either a new conjunction, or a failure of a conjunction that we thought was there. And it's always trying to live inside the data and use it to test the structure. One of the more exciting things that we discovered maybe almost a decade ago now was a mechanism for the autoregulation of this transcription factor POU.1. Which we knew had positive autoregulatory binding sites in its regulatory systems. And so we expected that when the cells ramped up its expression, it would be going through this transcriptional mechanism. And then I got a systems biologist into my lab who did live imaging, and we live imaged the levels of POU.1 transcription coming off a fluorescent reporter. And we found that the slope of fluorescence increases over time, which should have been the direct output of this transcriptional mechanism actually didn't change when we thought this factor was positively auto-regulating. So instead of that what was happening was that the cell cycles were getting longer. And there was such a strong expectation that this positively auto-regulated and that it should be doing it by increasing its own slope, but it didn't. And instead, it was working on the cell cycle length. And so how could that possibly make a difference? Well, the only way it could make a difference is if the protein was stable enough, so that the cell cycle dividing the amount by half was actually significant for its total level. And sure enough, it turned out that this transcription factor in this context is rock stable, you know, it lasts for days. And so in fact, it shows you that you can have an entire mechanism of differential regulation of the activities of different factors that collaborate together at different balances by playing with cell cycle versus protein half-life. And I mean, nobody had this idea before, that I was aware of. So here we had the strong expectation, it was totally violated by what we saw. And we had to come up with another explanation, which actually turns out to be true. And so that was an example of something that was quite exciting.


Martin   

So do you still remember how you had that insight, or someone in the lab had suddenly that insight that it had to do with this interplay with the cell cycle?


Ellen  

Well, no, I did. I mean, I looked at it, I said, you know, where is this up-regulation. And what is happening to the cells, and they're going so slowly. And then we realized at that point that we had to measure the half-life of the protein, because that's the only way that you could account for the buildup of the protein without increasing this. Somehow, it had to be reducing its degradation rate. And when you see the cell cycle, you know, every time you're watching a fluorescent reporter, the level gets cut in half, by the cells dividing the level per cell. And so you could see the baseline creeping up. But you could see that it was creeping up, basically, because of this elongated cell cycle time. You know, that was a place where the data talked to us, and we were willing to listen. And I think the willingness to listen, and to be excited when your expectation is violated, is really crucial.


Itai  

And I think that's beautifully put. You have a hypothesis, and if it's wrong, well, first of all, it takes a big person to admit that the evidence doesn't support the hypothesis. But let's say you even make it that far, there's this expectation that okay, you would now come up with a new hypothesis. But I think what you highlight is such an important point that there is this conversation with the data, that the new hypothesis is not just a random one that you just brainstormed, like you actually look at the data, and I like how you say that you inhabit the data. It's almost as though you're putting yourself in there with the cells and try to come up with what's going on.


Ellen  

Very, very much. I know that everybody knows the story about Albert Einstein, you know, imagining himself being a photon. But in a sense, when we look at data sets, that's what we should be doing. Even if we're looking at molecular biology data. If you are this transcription factor in the cell's nucleus, How the hell do you figure out which of the, you know million possible genomic sites you're going to go to in this particular cellular context? And we don't actually have a good answer for that. We have all kinds of algorithms for looking at preferred motifs. But they work very poorly to predict where transcription factors in mammalian cells actually are found binding. And so you know, this really helps you identify what I think are really major outstanding problems in one's field.


Itai  

I think it's really interesting how you say, when you say inhabit the date, I kind of imagine you focusing on a particular gene and imagining the world from that particular gene’s point of view. But then you say that, in contrast to the old days, where everything had to be hypothesis-focused, now, you have these genomic data sets. So I'm wondering how it is that you can sort of focus in on individual genes, given the wealth of data sets that can tell you everything, about all the genes. How do you do it?


Ellen  

Yeah, I think that that's actually a big challenge. And I think it's something that our field really hasn't come completely to grips with yet. You know, I have a bit of an advantage, because I've been a scientist for a long time and I've been working on T cell development for a long time. And so I started out at a time when very few of these genes were known. And so they have very strong personalities. To me, they all have backstories, you know, and I have a slightly easier time fitting new pieces into that structure. I think for younger people, it's really tough coming in right now. And there are these very well-intentioned tools like gene ontologies, that are supposed to help, but they actually make your blood run cold, how poorly they actually do to represent what these genes are actually involved in known systems. So I do think this is a terrible challenge. I really don't know how to cure this problem for people who are coming in, where they're just hit with, you know, 25,000 lines of Excel spreadsheet and your data to four decimal places, and it's tough. I think it's good to think of distinguishing features that make a cell a cell and to basically find these landmarks in one's datasets. And then classify into modules. I think the modularization is really important. What are identity features of the cell? What are functional programs? Submodules. What are for T cells, what are immunoreceptor selection modules, in terms of gene regulation. And then you start to recognize these clusters, and you can track the clusters, but they're functionally defined for the history of your cell type. So that that actually helps you keep track of them, regardless of what the gene ontology tool says that those genes are involved in, in some other cell type.


Martin   

So that sounds like one of the tools that you're using, is to just have a set of questions that you always ask.


Ellen  

Yeah, if they're not the question, they're there as the kind of the orientation questions like, Am I really in the room that I think I'm in? That kind of thing? And because sometimes you find that you're not. And sometimes you find some very surprising things about, for example, the way enhancers work, or the way T cell precursors differ from innate lymphoid cell precursors, with respect to their deployment of the exact same transcription factor, totally different places of the genome, guess what expression of this factor is not sufficient to predict what it's doing, you know. And so those orientation questions – is this thing, the sort of checkoff list. Is this really doing what I expect it to be doing? That's really important right up at the start. And then once that framework is established, you can ask the question that you thought you were intending to ask in the first place.


Martin   

Yeah, yeah. Yeah. You said something else a bit earlier on, that I also found very interesting. You said, I imagine that I'm the gene, how would I go about doing what I want to do? Do you have a different way of talking? When, when you're in that mode?


Ellen  

I think I use way too many metaphors all the time. So any number of different ways.


Martin

Yeah. And you know, you couldn't you couldn't write a scientific paper like that. Right. You could now write introduction, if you imagine being that transcription factor. It doesn't work in what we would call day science, right? But in night science, it's a fantastic way to kind of try to understand what's going on in the process you're studying.


Ellen  

Yeah, exactly. Exactly. And I think that this is really true. We all suppress our individual voices as much as we possibly can, in our scientific writing, which is too bad actually. You know, that was something that has changed. People used to actually be able to write as writers you in earlier days, and nowadays we have to sound extremely similar to each other. But I think, obviously one wants to be precise about what one is talking about. But these analogies, I think, help us understand where the prevailing structure falls short.


Martin   

Yeah, Itai and I actually recently reviewed a great book about how science works the knowledge machine by Michael Strevens. We really liked it, we think it really explains a great deal about how science works. And what he's saying is that this is the iron rule of science, and that is, if you want to decide who's right, the only thing you can argue with is data. So all you can bring into the discussion is data. And that's how we write papers. And that's how we give scientific talks. But then to actually get to that point where we have a story to tell, you know, they can do whatever you want, it's just in the papers that you have to take a step back and only argue with the data.


Ellen  

Right. And the thing is, the data give you the path to understanding something you didn't understand before, but they don't necessarily tell you what it is, there are a lot of different ways that it could be. And I think this question of climbing into the data, sort of imaginatively is sort of to harness the structures our brains are built with for very intuitive ways of learning about the world, you know, ever since we're little tiny babies, and trying to see if any of those intuitive modes of navigating, are showing you anything about the data structure, then, of course, the second you've articulated what the question is that you think is interesting and new, you obviously have to go and very rigorously design the experiments that are going to test that. And you have to be willing to pitch that idea out just the way you would pitch out any other idea. I mean, you have a very high bar for actually accepting a new idea. But if you haven't thought about what the new idea might explain, that isn't already explained, you will never be impelled to look for the new idea or to test it. So there is this dialogue back and forth all the time between the day and the night science, if you want to call it that. And that is important. The other thing is that being a scientist means constantly questioning everything, including your own pet structures. And I think one of the real disservices that our publicity does to the world, is to talk about what we've learned as though we're asserting it, rather than talking about it as something, which is actually the best we could do with numerous, numerous attempts to do better. And that we never stopped trying to do better in terms of explanations, we never stopped trying to do better in terms of challenging our own presuppositions. And the essence of being a scientist is this, you know, cherishing and curating the structure in order better to see where it might not fit, where it might have to be fixed or broken and rebuilt. So it's something that many people in their normal lives are not used to doing. And I don't think we do a very good job of explaining it to the public.


Itai  

We don't explain it. And it's probably in part because as you say, when we present it to the world, we have to do that in a robust fashion. And then the exploratory idea generation gets naturally hidden.


Ellen  

Yeah, although I think in a way COVID has been useful for this, because people have to watch in real time, all over the world, as scientists grapple with something that they really did not understand at first, except for some basic principles, and sort of watch the process of trying to learn and some people obviously have missed this entire dialogue. But others have been watching and i think that others out there may have been learning much more about science than they knew before, about the process, about the way you suggest ideas, the way you actually test it against data. And the way that makes a difference to the world. 


Martin   

Yeah, but initially, it was almost funny slash scary to see, you know, how people didn't understand this process of silence, just like you said, you know, like, there were people saying, you know, but but like, two months ago, the scientists were saying one thing, and now they're seeing another thing. So how can we trust these people? 


Ellen  

Right, yeah, well, yeah, exactly. Right. Some people learned that science actually takes sometimes as long as two whole months.


Itai  

So Ellen , keeping with this theme, I wanted to ask you, how do you distinguish any idea that you have as sort of a better than the frey idea, a great idea that you want to propose to test and since you mentioned Eric Davidson, I wanted to mention this story. So I was just starting my lab, and I had this interaction with Eric that, actually I came out very sort of traumatized by it. But in retrospect, I guess it's what I needed to hear. I explained to him an idea that I had. And he didn't like it very much. What he said was, he said, See this peach, you know, it takes a peach. He said, Well, you're talking about, you know, you're sort of idea, it's just the fuzz on the speech. And what I'm interested in is the peach. I want to know the peach, not the fuzz, you’re understanding? I was very traumatized. But he, he sent me straight, you have to focus on a better idea. So how do you do it? How do you know when you have an idea, it's okay. It might be an interesting idea. But it's not really what you want to devote the next two years of your life to?


Ellen  

Well, I think I try to generate lots of ideas, some of which are smaller than peach fuzz, some of which are tough, nasty skin, and some of which may lead to the sweet meat. I think that you don't know how far an idea is going to take you until you follow it a little ways. And you can't get too emotionally invested in it when it's still on the outside. Now, it may be that your goal is to define the topology of the surface of the piece, in which case, looking at the surface might be very important. I think, again, the issue in your conversation with Eric would have been, you know, what is the question? But honestly, you have to follow it a little ways to see how robust it is before you see how far it takes you into understanding a large new problem that you haven't been able to get at before. And one has to take kind of tentative steps bit by bit at the time. Now we as scientists nowadays have a big problem, which is that we have to constantly oversell our work and our ideas. And we have to constantly describe it, say to reviewers and funding agencies, as though every experiment that we do is going to solve the greatest problem in the world and that it's absolutely guaranteed to come out this way. And we and the reviewers know this isn't true. I don't actually understand how we allowed it to get to this point. I mean, it certainly isn't true in the practice of being a scientist. But I think the fact is that you can come up with a lot of different thoughts and a lot of different avenues of investigation, which will be transformative sometime down the line, but you have to be allowed to develop them to get them to that point. I deplore the fact that we are being pushed into overclaiming constantly, when work is at an early stage, that is the stage when you actually have the chance of starting something that will be transformative later on. But you have to let it grow and get stronger first, and get it debugged first to a different metaphor. So this is a problem. And sometimes you have an idea, which turns out to be dead wrong from the start, that's very easy. Sometimes you get an idea, which is good for a small local problem, but it really doesn't have any throw weight when you go to other problems. And then sometimes you come up with something which you then recognize as being a mechanism that's involved in a lot of phenomena that you'd hadn't had an explanation for before. And that's really exciting. But you have to get to the point of testing that first.


Itai  

And you said at one point, I don't know how we let it get to this point. Do you mean that? 


Ellen  

We, as the scientific community, I think we've collapsed. We're doing very poorly at this point. 


Itai  

I think I see it too, that for some reason, sometimes sort of being playful, which many of us recognize as kind of like the first steps of project is being seen as a liability is being seen as something that suggests bad science.


Ellen  

Yeah. And we write papers backwards, right? I mean, we write them as though we didn't know what we know, at the end, which was actually the last thing we learned


Itai  

In this writing of papers backwards, I think in many times it can confuse the next generation of scientists.


Ellen  

It does. It does. Yeah, and the courses that I've taught, often when we dealt with scientific literature, often we had to go back to papers from more than 10 years ago, in order to come up with anyone who even articulated any fraction of the actual logic process involved in the research. So the writing of papers has actually started to obscure the scientific process. I think this is a problem. And I think that also as these younger people are now grant reviewers and paper reviewers  but the template isn't necessarily accurate. So I think that this is sort of screwing up our ability to communicate properly what we do, but nothing that I'm saying should go against the idea of needing to establish things with extremely rigorous and extremely searching critical experiments and tightly controlled measurements. It's just that you have to first have an idea of what's worth measuring.


Martin   

Yeah, you know, it's almost like we should all publish a supplement to any paper that we publish, where we describe the process that led to a hypothesis that we then stayed in the introduction of the paper.


Ellen  

I think that's a wonderful idea, I wish we were allowed to do it. It's such a shame because publishing became so distorted by the need to fit what used to be five papers worth of data into one paper, and then to have draconian page limits or character counts, so that you were actually forced to cut all the logic out of your papers in order simply to fit it into the character count. I think that that's like the last hurrah of the print journal phase. And now that we have online publication, more and more predominantly, maybe – hah, fat chance, this is a cynical element  –  but maybe some of this could be relaxed a little bit. 


Itai  

You know, I never thought about that. What if the rise of these tabloids Nature and Science with their, as you say, draconian page limits, has led more and more to the hiding of the underlying logic and more of just clear presentation of the results, that’s it.Martin   

? I'm actually afraid that the habits of social media would go into this bad direction, right, you know, no tweets, how many? How many letters are you allowed? Right? And so you have to summarize a paper in that amount of space, that's even a lot worse than three or four pages of Nature.


Ellen  

Yeah, you're absolutely right. I think that this is a serious problem. And one thing is, I feel tremendously grateful that I've been a scientist during the decades that I've been a scientist. But I get very worried about whether I'm able to leave the world to my trainees, as such an exciting place to do science, as it's been, during the decades that I've been one. t's been such an incredibly wonderful profession to be in these years, just absolute joy. And I hope that our trainees are going to be able to have a large piece of that joy going forward 20-30 years from now, but I don't actually know what form it's going to take.


Martin   

Yeah, but I think you're absolutely right. That's one of our most important tasks to prepare the next generation of scientists for a creative and exciting career. So do you have any anything specific that you do to mentor your students, especially for the creative process in science,


Ellen  

I don't know, you know, I do my best, I've never had a large lab, the people who have come to work in my lab, many of them have been just absolutely wonderful. Each one is totally different from the others. And so there isn't a one size fits all, I think that I tried to get to know their marvelous qualities. And I try my best, which is obviously imperfect, to create an environment in which they can build on their strengths, in a kind of a dialogue with me, but it's, every person is different. And they prefer different kinds of ideas. And they prefer to follow different kinds of research. And so thing number one is obviously to allow them to develop the highest success with their own preferred talents, and to follow their own tastes, and not try to mold them in some way that is not necessarily what they want. But they are facing a very different world than what I came up in. And so I do worry about, you know, where they're going to find the breathing space for the kind of creative thinking and exploratory science that I was lucky enough to enjoy.


Itai  

And if I could follow up on that, what is a typical meeting by a graduate student look like with you?


Ellen  

Sometimes it's more than three hours of crawling through the data, and just brainstorming, throwing ideas back and forth. I mean, it's all different kinds of things. I don't have a disciplined way of having, you know, sort of regular 30 minute meetings with each of my people. Okay, times over, time for the next meeting. That's not the way we do things.


Itai  

And do you have a particular way to handle when they're frustrated? Perhaps the project's not going their way? I feel like psychology is sometimes...


Ellen  

Yeah. And we're all amateurs, right? Look, the thing is that I'm actually really optimistic about science. And I think that when an experiment just isn't working, I just tell them, Look, this is what science is like. It's an experiment that you haven't had worked before. It can take a while to get it to work. There are a lot of variables. Don't worry about it. Don't panic. You're not bad. It's just a process. You're going to debug it. Just be systematic about it. It'll work. When it comes to actually data that doesn't fit with what they expected, I tried to set, let them see, this could be really, really, really interesting. And so, we try to actually get into it at that point. What else did you see in these experiments that came out that way? What else could that be? Meaning what could that be telling us about system? Sometimes it's exactly these times that we see these observations, they start out as being depressing. Oh, I didn't get the result I expected to get well, let's look at it. Oh, my God, this is even more interesting. This is much more interesting. Let's take a look. This is sometimes how we make discoveries. So I think that if you actually reproducibly get a result, which isn't what you expected. Sometimes it is a whole heck of a lot better than what you expected in terms of breaking new ground scientifically.


Martin   

Yeah. Like discovery, by definition has to be unexpected. But it's really interesting to me that you said you don't have a systematic way of how you interact with the students. But what I understood was that you actually climb into the data together with your students. Right. And I tried to go on a discovery tour together.


Ellen  

I tried to. I hope they feel that I'm doing that with them. Yeah, I mean, that's, that's very much what I like to do. You know, I mean, in my normal life, when we don't have COVID, I have a busy schedule, and I travel, probably too much. And so the students and trainees don't get to see as much of me as perhaps they ought to.


Martin   

So is there a typical way of how a project in your lab unfolds? I mean, do you usually start with a hypothesis and then just realize, Oh, it's not working that way, it's gonna be something else? Or do you just start by producing some data that you think is going to contain interesting phenomena? And then just look at it? 


Ellen  

Well, I guess a lot of the projects start in different ways. But I'd say that a lot of times, you know, based on things that we've been working on for a while, there will be actually a hypothesis set up as a question where either answer is interesting. So is this working such in such a way ? Or is it working such in such a way? So you try to come up with experiments that would distinguish between those two answers? And so that way, it's a lot less likely that you'll just feel that you failed, you actually find out something, those projects generate usually a lot of results, which are sometimes quite surprising. For example, the use of an enhancer for gene regulation, is it being used only to kickstart the gene expression at a particular stage? Or does it have a continuing additive effect on gene expression throughout all subsequent levels? And so in order to pose the question that way you realize what sort of problems, what sorts of technologies you're going to need to deploy in order to measure things with the accuracy. But then in the process, you find that all kinds of things even know you look at a transcription factor perturbation genomic data set, and you see clusters of genes being affected that you didn't expect to see. And so, this becomes a hypothesis generating mechanism in itself. So, I think that things often end up in different directions than the way they started. But they usually did start with a question. Is it more like this? Or is it more like that?


Martin   

Yeah. You said there's different ways of how projects unfold. But to me, it sounds like whatever it is, you always approach it with a very open mind. Right? So, either you just have a question, or you have multiple hypotheses. And so, I think that's a great way of making discoveries. Right, but just being open to what the data is trying to tell you.


Ellen  

Yeah, I definitely really dislike the kind of science where you have to do an experiment, needing to get a certain result in order to be a success. I think I actually shy away from that. And I think that's caused me to shy away from clinically associated research altogether, because the benchmark is so stark, you're not saying is it this way or that way? You're saying does this work better? Yeah. And there's only one possible answer, or else you're a failure. And to me, that is not an appealing kind of science to do, although, obviously it's very useful for the public service. But the kind of science I want to do is exploratory, where as long as you can formulate the alternative answers in a clear and cogent, falsifiable way, then either of the answers could be extremely productive.


Itai  

Right. It's not as though you're trying to make something better or worse. You're trying to figure out reality you're trying to explore and understand. And you know, in thisrespect, one important aspect is the availability of new tools, right? And over the  years, it's just been amazing how powerful new tools keep arriving. 


Ellen  

It’s exhilarating. And you can finally ask questions you could not possibly ask before, the way you can actually image live single cells going through developmental decisions in real time. You can play with them, you can perturb them, you can watch what they're doing, you could look at their whole transcriptomes. It revolutionizes what you can say about them, how you can understand how they make decisions. It's extraordinary.


Itai  

Yeah. And I think you know, it's extraordinary for you. And I think that's very admirable. I think for many researchers, it's maybe overwhelming, and it's really great. You're, I think, your attitude,


Ellen  

if you live in the questions, if you live in the questions, the questions are always bigger than the experiments that you currently have the technology to do. So someday, you know, 5-10 years down the line. There's a technology out there, suddenly, it can answer your questions. Usually, it comes in from a totally different field that can come in from neuroscience, it can come in from physics, but suddenly, it's there. And you can use it to get that question that you've had for maybe 20 years. But you can do it now. And you can pounce, it is a joy. It always has bugs working it out at the first place. But it's incredibly exciting. And, you know, it makes you feel that you never ever, ever want to stop doing this. It’s endlessly marvelous.


Itai  

I love it. And he knows you forget who said it. But I think the quote is, theories come and go. But the embryo stays. 


Ellen

Yes, yes, exactly. 


Itai

I think from your perspective, it's these methods, they come and go ever new, ever more powerful. But the questions remain.


Ellen  

You try to get all the juice out of each method, and then you realize that you can connect it with something new and answer a whole new kind of question. So yeah, it's a marvelous field to be in. If one is lucky, to have the support, and to have brilliant young people come and work on the problems. I really get excited that the young people who are setting up their own labs and finding marvelous discoveries, and I just hope that they're going to keep on having these. These make me feel good about the future. And I really, really hope that they get the support through their careers, to be creative, and to think widely about the kinds of questions that are interesting that I've been allowed to do in my career.


Martin   

Yeah, I think that's actually a wonderful way to end this podcast, you know, by emphasizing how much joy you have in what you're doing and how much you're hoping to contribute to the next generation of scientists to experience the same kind of joy of livingthe questions.


Ellen  

Absolutely. Thank you so much. That's a perfect paraphrase.


Itai  

Yeah. Thank you, Ellen. I think we introduced you as someone with infectious energy for science. And I think that's exactly what we have heard. 


Ellen  

Thank you so much. It's been a pleasure.