Night Science
Where do ideas come from? In each episode, scientists Itai Yanai and Martin Lercher explore science's creative side with a leading colleague. New episodes come out every second Monday.
Night Science
5 | Sarah Teichmann’s artist colony of scientists
In this episode, Itai and Martin talk to Sarah Teichman, Head of Cellular Genetics at the Wellcome Trust Sanger Institute and Director of Research in the Cavendish Laboratory in Cambridge, England. In her creative research, Sarah’s thoughts constantly switch between her native languages – bioinformatics and genomics – and foreign languages, such as chemistry and physics. Sarah talks about storytelling vs. modeling when interpreting data, and discusses hard vs. soft hypotheses.
Sarah is interested in global principles of protein interactions and gene expression, focusing her research on genomics and immunity. She is an EMBO member and a fellow of the Royal Society and the Academy of Medical Sciences. Sarah received numerous prizes, including the Lister Prize, Biochemical Society Colworth Medal, Royal Society Crick Lecture, and EMBO Gold Medal.
For more information on Night Science, visit www.night-science.org .
Sarah: And when you're switching between your native languages, you don't really notice actually. And so that's a little bit probably like bioinformatics and genomics maybe for me, or kind of my native languages. Maybe chemistry, physics, maybe a little bit like foreign languages, you know, I've learned them, but they're not so deep. I have to kind of switch maybe more consciously.
Martin: Welcome to the Night Science Podcast, where we explore the untold story of the scientific creative process.
Martin: We are your hosts.
Itai: I'm Itai Yanai.
Martin: And I am Martin Lercher.
Itai: Sure.
Martin: Welcome, Sarah Teichmann.
Sarah: Thank you, Martin.
Sarah: Thank you, Itai, for having me here.
Sarah: I think it's really fascinating, and I just wanna say thank you to both of you for kicking off this initiative.
Martin: So just as a brief introduction, Sarah earned her undergraduate degree in biochemistry at Trinity College in Cambridge.
Martin: In , Sarah became a principal investigator of the Wellcome Trust Sanger Institute and the European Bioinformatics Institute.
Martin: Since , she is serving as the head of cellular genetics at the Wellcome Trust Sanger Institute, and Sarah received many honors.
Martin: And so I can only give a few examples.
Martin: She won the gold medal of the European Molecular Biology Organization, and she has been elected a fellow of the Royal Society of the UK.
Itai: Yes, and in her research, Sarah really represents a kind of new breed of scientists that work at the interface between the computational and the experimental side of biology.
Itai: Her group at the Sanger Institute is elucidating the general principles of gene expression and complex assembly of proteins.
Itai: And in particular, over the last few years, Sarah's lab has emerged as a true leader of single cell genomics.
Itai: Personally, I've known Sarah for many years, and recently I had the pleasure of organizing a conference together with Sarah, which was a lot of fun.
Itai: On her lab website, the motto of the lab is, be bold, be brilliant, be kind.
Itai: Sarah, it's so great that you're joining us today.
Sarah: Thank you, Itai.
Sarah: It's a pleasure and I look forward to our discussions.
Martin: Yes, I think it is actually a great motto.
Martin: So that's probably what we all strive for.
Martin: So to get started with our Night Science discussion, Sarah, are you aware of any aspects of your training as a scientist that maybe influenced your development of becoming creative?
Sarah: You know, as scientists, we go through a lot of rigorous training and we don't really have an explicit training of creativity.
Sarah: So it's different from, let's say, art school or journalism school or creative writing kind of courses.
Sarah: In my case, I would say it started off with the fact that I read the Natural Science Tripos at Cambridge as an undergraduate.
Sarah: And the benefit of that undergraduate degree, when I look back is that you're not restricted to one science, you're exposed to lots of different sciences.
Sarah: And so that meant that I had training in math, physics, chemistry and biology, you know, for the first few years of my degree.
Sarah: And that meant that me as well as our, you know, all my peers were exposed to many different schools of scientific thought.
Sarah: So math is very different from molecular biology.
Sarah: And it's really the benefit of having access to what I look at different toolboxes and different scientific cultures that has maybe contributed to the way I do science, as well as then later working in research labs where you have role models and you see people that are creative and you see how also the interactions between different scientists can lead to creativity.
Sarah: But I think that the interdisciplinary training is really a good thing.
Itai: Do you feel that you can channel the thinking of a biologist differently than maybe at other times when you feel you have to channel the thinking of a chemist?
Itai: Is that what you mean by that?
Sarah: Yeah, that's what I mean.
Sarah: There's the data science, which is more statistical computational toolbox that we have and that we use.
Sarah: But then there's also the biology.
Sarah: And then there's in the case of the protein biophysics and so on, it's chemistry really and physics.
Sarah: And so they are different ways of thinking and they're different tool kits.
Martin: And do you sometimes consciously switch between those different ways of thinking?
Martin: Or is it something that's somehow integrated for you?
Sarah: I think, you know, it's a little bit like speaking different languages.
Sarah: And there are some languages that are your mother tongue or that you speak natively.
Sarah: Like for me, it's German and English.
Sarah: And then there are some languages that you know, but you know them as a foreign language.
Sarah: So for me, that's French and Italian.
Sarah: And when you're switching between your native languages, you don't really notice actually.
Sarah: And so that's a little bit probably like bioinformatics and genomics maybe for me, or kind of my native languages.
Sarah: Maybe chemistry, physics, maybe a little bit like foreign languages, you know, I've learned them, but they're not so deep.
Sarah: I have to kind of switch maybe more consciously.
Sarah: That's a little bit how I think of it.
Sarah: So we are switching more or less consciously, I think all the time in our work.
Martin: Yeah.
Martin: Another thing that you mentioned just in passing was that then after your first degree, you experienced different PIs and their way of being creative.
Martin: Is there any particular example that's important to you?
Sarah: My PhD mentor, Cyrus, I would say he had a really different way of being creative from Janet Thornton, my postdoc mentor.
Sarah: And those interactions are quite formative actually.
Sarah: And you observe your mentors and then copy them, I guess, to some extent, or take some things away.
Sarah: And Cyrus used almost storytelling as a way of stitching together different scientific observations, whereas Janet is much more statistical and computational.
Sarah: And I guess, but there's not a right or wrong.
Sarah: Both are creative kind of ways of interrogating data, for sure, but they're very different.
Sarah: And I guess I've kind of taken a little bit from both.
Itai: That's so interesting.
Itai: And you mentioned that Cyrus had this technique of trying to tell a story.
Itai: Can you tell us more about that?
Sarah: When we were looking at data together, he always tried to interpret and explain the data in a certain narrative.
Sarah: And so Cyrus would interpret, you know, would look at the data, and then he'd kind of intuit that this could mean that a large gene family is more likely to duplicate and lead to this power law decay of protein family sizes, let's say.
Sarah: Whereas Janet would, I mean, I never discussed this same data with Janet, but she would more like say, okay, let's fit different functions to it and see, and let's model it.
Sarah: He would use words much more to make up a story that could explain the data, which is then a kind of hypothesis formation, and you can then go and make your model.
Martin: Yeah, but then if you fit a function, then the function that's the best fit also will lead to some hypothesis.
Martin: It's really two very different ways of approaching the same kind of problem.
Itai: So to keep on this thread of the creative mode, when you're meeting with a team member in your lab and you're talking about a particular project, do you actively make a distinction between the sort of executive mode of, okay, we're going to test this, this is the controls, the power of the analysis that we need, and on the other side, the creative mode, trying to come up with the idea to be tested in the first place.
Sarah: I think in our daily life, research is % perspiration and only % inspiration.
Itai: Without that inspiration, we're nothing.
Sarah: That's right.
Sarah: Without the inspiration, thinking about it, we probably don't make enough time for the inspiration and for explicitly carving out time for creative thinking.
Sarah: It more kind of pops up in the gaps between routine work, in the discussions that you have over tea, the corridor discussions, or driving or cycling to and from work with colleagues and so on.
Sarah: Maybe we should make more time for explicit creative thinking and brainstorming.
Martin: I totally agree.
Martin: But probably this time in between is kind of the natural habitat for Night Science.
Martin: And maybe that's why Jacob actually called it Night Science, right?
Martin: Because it's not usually what you do during the day, during your working hours, but like you say, in between.
Sarah: Yeah, I completely agree.
Sarah: I completely agree.
Sarah: And it's either sort of musing yourself and turning things over in your head, or it's discussing within the group or with collaborators or colleagues.
Sarah: I mean, sometimes even people who aren't scientists when you're explaining your work.
Martin: So one thing you mentioned earlier on was that these different disciplines are almost like different languages, right?
Martin: And that now you have like your native languages and then you have other languages that you're not quite as fluent in, but that you're still able to speak to some extent.
Martin: But when you talk about science, do you also use a different language when you are discussing things creatively from when you are talking more about day science?
Sarah: I think there isn't a conscious distinction, but certainly the language that we use when we're kind of having relaxed conversations aren't the sort of formal language of manuscripts and papers, for sure.
Martin: One thing that Itai and I like doing is to use anthropomorphisms.
Martin: Is that also something that you do?
Sarah: I think anthropomorphisms or analogies.
Sarah: I use the analogy of the foreign language and the native language and bilingual.
Sarah: I guess that's a really powerful way of explaining things to each other, even.
Sarah: And, yes, Muz (Muzlifah) Haniffa is a master at this, and she's a close collaborator of mine.
Sarah: She's a real master of the analogy.
Sarah: Sometimes anthropomorphic, sometimes not, but definitely it can be really powerful for triggering creative modes of thinking.
Martin: So beyond using a slightly different language when you talk in a more creative mode, do you have any conscious methods you use when you try to come up with ideas?
Sarah: If I have an idea, certainly running it by other people is something that I do a lot, and that's incredibly helpful.
Sarah: And, you know, that's really where the value of the group members, the colleagues comes in, and that can then trigger, you know, a further development that can be even better or a sharpening of the idea, if it starts off as a sort of fuzzy concept.
Sarah: And that's the interaction and the collaboration is also a lot of fun, you know, when you've got really great colleagues and it breeds a lot of creativity.
Sarah: Reaching across borders to collaborate across disciplines, I think, can be such a fantastic way of forging new ground or making progress.
Sarah: And I mean, the human cell atlas is an example of that, I think.
Sarah: It's a project that aims to map all the tissues in the human body using high throughput single cell genomics and spatial methods.
Sarah: And that all has to be stitched together with computation.
Sarah: And so you've got a community there where you've got clinicians, the surgeons and the people who are really close to the clinical samples from the human body.
Sarah: But then you go all the way over through the biomedical experts, the genome biologists, technology developers, and then the computational community, which both Aviv and I come from, of course, with the statistical computational and now sort of machine learning or the deep learning that's come more recently.
Sarah: And it's that interdisciplinarity that to me is always an important breeding ground for creativity and is super exciting to me.
Itai: Sarah, you're in this field where you generate so much data and you recognize how much can be explored and discovered in this data.
Itai: Do you ever notice a tension between some sort of scientists that may feel that science should be only hypothesis driven and then maybe another subfield may recognize that the data science is working precisely in this exploratory way?
Sarah: Well, I think there is a perceived tension that there's somehow a difference between the manner and the value of the conventional hypothesis driven sort of mechanistic studies as one molecule at a time and what does it do and dissecting each element of a signal transduction cascade or something like that.
Sarah: And the large scale high throughput genomics research that we are in or computational research.
Sarah: But personally, I think that the distinction is actually misleading because I'd be interested to know also what you guys both think.
Sarah: But to me, actually, both types of science include both hypotheses and data generation and interpretation.
Sarah: And you're not just generating genomics data in a vacuum.
Sarah: You know, when we decide to study the placenta and the decidua, we had a very specific reason that we were taking that tissue.
Sarah: And the reason was that we wanted to understand maternal tolerance of paternal antigens.
Sarah: So you go in with a hypothesis, you generate the data and then you interpret it.
Sarah: And when you do the data interpretation, it's not like machine learning or AI is completely hypothesis free.
Sarah: But again, the way you construct the models, the way you apply them, you have some kind of mental model of the data and the questions that you're asking.
Sarah: It may be a mathematical, mechanistic model, but nevertheless, there's a reason why you use certain approaches and not others and how you optimize parameters.
Sarah: So to me, there's a constant interplay between hypotheses, data generation, data interpretation.
Sarah: It just goes round and round.
Sarah: And it doesn't really matter whether it's the conventional biochemistry, molecular biology or whether it's genomics.
Martin: We actually call that the data hypothesis conversation.
Martin: In order to generate a specific hypothesis, you have to look at some data, right?
Martin: It doesn't fall out of the sky.
Sarah: Exactly.
Martin: And you cannot look at any data without having a lot of theoretical background in the back of your head or even built into the methods that you use to analyze the data.
Martin: In our view, the distinction is that whether you are testing a specific hypothesis with the data, so you get a yes or no answer or a p-value, or whether you're using all that hypothetical background to just look at the data and let the data tell you what the precise hypothesis should be.
Sarah: Yeah, I see it more as a quantitative distinction rather than a qualitative distinction.
Sarah: There are shades of, it's not like white or black, hypothesis driven or exploratory.
Sarah: There are many different shades in the middle.
Sarah: Let's say you do a clustering, for instance, and yes, I agree, you can do it in a completely hypothesis-free manner, but then you have to interpret those at the end of the day.
Sarah: You're not going to get very far with just publishing a bunch of clusters and numbering them.
Sarah: You have to then take it one step further and interpret that.
Sarah: That's where the hypotheses come in and the prior knowledge.
Itai: I can see what you mean with the shades of gray.
Itai: Sometimes I think of hard hypotheses and soft hypotheses.
Itai: A hard hypothesis is, if I compare these two cell types, this gene is going to be differentially expressed.
Itai: It's an extremely well-designed hypothesis to be tested.
Itai: A soft hypothesis is more, I bet you that there would be subtypes in this particular cell type.
Itai: I don't know what they are, but they should exist.
Itai: Let's go look if they exist.
Sarah: Maybe that's a good way of thinking about it, actually.
Sarah: And maybe it's that we tend to work more with soft hypotheses rather than hard hypotheses in general.
Sarah: But then if you want to go in and validate one very specific cell type or molecule, you're going in with a hard hypothesis at the end of the day.
Sarah: But I agree that data science in general is more the soft hypothesis.
Martin: Yeah, that makes sense.
Martin: So Sarah, I'm sure that you also try to help the people you work with, especially the students, to develop that creativity for themselves.
Sarah: I think there are two main elements really at the end of the day.
Sarah: So one is what we discussed before about interdisciplinarity and collaboration with people across disciplines.
Sarah: This afternoon I gave my group a sort of sermon, a talking to kind of in the group meeting about how important it is that they are open with each other, that they help each other, that they collaborate with each other, and that that's to everyone's benefit in the end.
Sarah: Within my own group, there are about people and it's totally interdisciplinary.
Sarah: So you've got clinically trained PhD students all the way to mathematicians.
Sarah: And so it's really important that they learn to talk to each other and engage with each other and respect each other and take each other seriously, even if it's an effort to kind of talk to people from a different discipline.
Sarah: But then the other part of it is, I think, giving them the confidence, the self-confidence that they can go and be independent scientists in their own right and have their own creative thoughts and follow their own projects and so on.
Sarah: Because I think self-confidence does have a link to creativity somehow.
Sarah: To give them that self-confidence that they can view themselves as really independent creative scientists or productive scientists, they can go talk to collaborators independently, so either within the institute or in other institutes and so on, that they are kind of a valuable independent scientist in their own right.
Sarah: They can come up with ideas and follow them.
Sarah: I would say that was my philosophy of mentoring, really fostering and understanding interdisciplinarity, but also encouraging independence and self-confidence in science.
Martin: So that is the be bold in your lab's motto?
Itai: You know, Sarah, that reminds me, I did a post-doc with Marc Kirschner at Harvard.
Itai: There was a visitor and he said to Marc, you know, Marc, I run my lab like an army.
Itai: You seem to run your lab like an artist colony.
Itai: And I think from what you described, your lab sounds like an artist colony.
Sarah: It's probably more like artist colony.
Sarah: I love the analogy.
Itai: I love it.
Itai: What is your advice for young scientists?
Itai: I mean, in particular, the ones that join your lab and are seeking a project, what is your advice for them on choosing a direction?
Sarah: Well, often people have something that makes them their skin tingle.
Sarah: I'm a believer in that skin tingling and the fun, also the motivation that people have for certain projects.
Sarah: And so I'm pretty flexible in negotiating projects that people are excited about and that fit into the framework of the lab and what the major themes and thrusts of the lab at the moment.
Martin: So yeah, the fun and the excitement of doing research, I think, is a crucial driving force without which it's really hard to get over the periods when it's not going the way you expected it to go.
Sarah: Absolutely, absolutely.
Sarah: And colleagues can be so important, you know, in keeping spirits up and keeping motivation going and so on.
Martin: Well, Sarah, thank you very much.
Martin: You gave us a lot of food for thought.
Martin: I have to say one thing I particularly liked was the image of the British artist colony where you discuss new ideas over a cup of tea.
Martin: No, it really reminds me of my postdoc in Bath with Laurence Hurst.
Martin: And so like every day he would come over to me and say, you know, what about a cup of tea?
Martin: And then we would talk science.
Martin: That was fantastic.
Martin: So it really reminded me of that.
Martin: Well, Sarah, thank you very much.
Martin: This was wonderful.
Sarah: Thank you.
Sarah: It was great to talk to you.
Sarah: It was a lot of fun.
Itai: It really was.
Itai: Thank you, Sarah.