Studying the Symphony of Cells

Featuring Aviv Regev, Executive Vice President, Genentech Research and Early Development.

Cells are the basic unit of life, with a remarkable ability to sense their environment, process information and adapt accordingly. With 37.2 trillion cells in the human body, there remains plenty for scientists to discover about how this symphony of cells is organized and works together. In this special episode, producer Wellington Bowler chats with Aviv Regev, Executive Vice President, Genentech Research and Early Development, to discuss her vision for the roles that single cell genomics and computational biology can play in not only forwarding our basic understanding of biology, but also in our ability to generate new insights about disease and ultimately develop new medicines for patients.

If you would prefer to read a transcript of this episode, please click here.



SUBSCRIBE BELOW TO CATCH EACH EPISODE

If you want to learn more about the groundbreaking science happening in our labs, click here. To learn more about the jobs in our research and early development group, click here.





Transcript of Two Scientists Walk Into A Bar: "Studying the Symphony of Cells” with Aviv Regev

 Hi, everyone. Two Scientists Walk into a Bar is back. The show for science geeks and the people who love them. I'm Wellington, the producer of the show. When we last met, we told you Jane had to take last call, but little did we know that it was going to be everyone's last call for a while, right? This year has been tough and like everyone else, we've been busy and stressed trying to live life during a pandemic. Wow. But we've really missed these conversations and we thought, OK, we need to do this however we can.Two scientists sitting for a video chat? Two scientists on a nice phone call, or maybe like in this case, one scientist and a Wellington. But I wouldn't do it alone, naturally, as we talked about, it takes an entire team to produce these shows. So, I am delighted to introduce you to my co-producer, Stephanie. Thanks for stepping on this side of the microphone, Stephanie. 

Stephanie: Hi, Wellington. Thank you. I have really missed these chats.

Wellington: Me too. Now, she would be the last person to tell you this, but Stephanie actually has her PhD in biology and so she's our resident science expert as we research and prep for our guests. Stephanie, who are we talking to today and why is her work interesting?

Stephanie: As our team gets back into the swing of things and we get ready for a new season, we had the opportunity to chat with Aviv Regev. She's a computational and systems biologist and I'll let her explain exactly what that means. Just to set the stage a little bit, though, I think what we’ll hear is a different way of thinking about biology, at least different from, you know, what maybe people are used to hearing. Aviv asks questions in a different way and at a different scale. And ultimately that's going to bring new insights into biology and disease.

Wellington: Awesome! Let's do it.


Wellington: Welcome to the show. 

Aviv: Thank you. 

Wellington: We're two scientists in a bar – well now it's one scientist and Wellington in a park. What's your take on COVID-19? 

Aviv:  I think what's interesting about COVID-19 for a scientist is that it's actually very rare for scientists to be posed with a completely new disease that didn't exist before. Usually when you start working on a problem, especially in disease biology, there is some body of knowledge of the disease, at least on its pathology or how it develops. There's exceptions to that, the most notable one being HIV. But that's been a while ago, HIV, several decades ago. 

And so this is an example of basically almost entirely data-driven biology. You can't just go to what we know before and use the model of how we understand biology before and apply it, you have to learn the disease afresh from data. I think that's difficult for people. I think for scientists, it's maybe a little less difficult because we're used to working on the unknown. But people are used to receive from scientists and from medical doctors more definitive answers. 

Wellington:  Let's back up, as Jane likes to say...You're known for single cell genomics. You're known for tissue biology, bioinformatics and much, much more. That's how others describe you. How do you describe yourself? 

Aviv:  I'm clearly a curious person. Everyone knows that, I think, at the end of the day. I used to say I'm a computational and systems biologist, not just because I develop and use computational tools in biology, but because I like to think of biology in the same way that computer scientists think. And I like to think about biology systematically. But if I had to really choose the piece of biology that I feel most identified with personally is cells – I love cells. 

I love thinking about cells and how cells work. I think of them as computational devices. So, I guess that's a little different than maybe what some other people think about them. And I always try to think about ways of understanding them better and dissecting them better and all the tools and methodologies, including single cell genomics and the computational tools and all sorts of other things that I built over the years. They've always been in service of just figuring out how cells work. 

Wellington:  Tell me more about the cell. What is so fascinating and what do we know and what have we learned? 

Aviv: Yeah. So first of all, cells are the basic unit of life. It's not DNA or the genes because DNA and genes on their own are not really life, but cells are the basic unit. You cannot reduce below a cell and still be a living organism. What's remarkable about cells is that they have this unbelievable ability to constantly sense their environment, process the information that they get about the external state and their internal state and shift themselves accordingly. And they do this through these elaborate molecular networks of genes and gene products, proteins and metabolites and so on, that interact with each other and change each other. 

Now, in unicellular organisms, the cell is alone and it interacts with its environment. In the multicellular organism it's even more remarkable than that because in the tissue you have multiple different kinds of cells. You go to a mode of division of labor between the cells and they have to balance each other out. In this ongoing situation is what we call homeostasis, where it's kind of the more things change, the more they stay the same. It's this unbelievable ability to respond to an ever-shifting environment and calibrate yourself accordingly so that you can maintain the tissue and the organism in a functional state. That's something very special. When it starts breaking down, that's when we get diseases. 

Wellington:  How are these cells different from each other and how many kinds are we talking about – millions, trillions? 

Aviv:  Thirty-seven point two trillion, to be exact. But they are not each a completely different snowflake. Biologists have been busy categorizing them since the 1600s when cells were first discovered under the microscope and definitely for the last 150 years or so. And you can categorize them in different ways. You can categorize them based on their structures, what they look like, their functions – which is what they do – where they're located in your body. 

And in the last few decades, with the rise of molecular biology, we basically categorize them by their molecules, the genes that they express, the proteins that they have, where their molecules are in the cell and so on. And this categorization is important for us because cells of different categories do different things and behave in different ways and perform different functions in the body. And at the crude level, if you went on Wikipedia, you would get a few hundreds of different kinds of cells. And I'm sure everyone has heard of those. You know, you have your neurons and you have your immune cells. Inside your immune cells, your T cells in your cells, and your dendritic cells and so on. 

Every tissue and every organ is made of this beautiful, well-organized tapestry of cells. And they make this beautiful symphony of function. And most of them or, we believe that for a lot of them, we don't yet know them. That's part of figuring out biology.


Wellington:  OK, Stephanie, I know you remind me that you're not at the bench anymore, you retired your pipettes, but help me sort this out. Aviv keeps talking about cells as the basic unit of life and I always thought that was DNA, where does DNA fit into this?

Stephanie:  Yeah, it's been a little while, but let me try to connect the dots. So, you can study biology at different levels. And DNA encodes the instructions for life, but by itself, it doesn't really do anything. It's like a book sitting on a shelf. It has a ton of information, but it has to be acted on. Someone has to pick it up and open it and read it. So, in this case, the DNA sits inside the cell and the cell has machinery that reads and transcribes and translates it, and then the cell can do stuff. And that's why we say the cell is the basic unit of life. It's really cool when you think about it, although I guess I'm just biased because I'm oriented the same way that Aviv is. I also love cells – I’m a cell biologist by training. It's just really how I think about biology.



Wellington:  So, I guess my question is, with 40 trillion cells, what do you learn by studying one and how one is communicating with the ones around it? And at what point, I guess I'm used to thinking things as either in pathways or systems or organs. But this is a new way of thinking about it. 

Aviv:  You don't actually study one. You study a lot. It's just that as you look at a lot of them, you look at each individual one rather than only at their average or the aggregate. So, if you think about it statistically, instead of looking, for example, at the mean of the distribution, you can look at all of the distribution or you can sample a lot from the distribution. So, you get many of their correct characteristics. You get the mean and you get the second moment, which would be the variance, and you get a third moment and so on. You can look at many, many of the features. 

And so, all of the cells of one type are not all identical to each other. They have these subtle variations between them. Some of these variations is because biology cannot actually reach this perfect precision where everything looks exactly the same. It actually tolerates this variation or this noise. But some of this variation is because they are not actually all exactly the same. They’re of the same overall category, but there might be subcategories. Or they’re of the same overall category, but they're not all experiencing life in exactly the same way right now. This cell might be in the middle of cell division, whereas that cell is currently quiescent, even though they're both of the same type. So, this variation between the different cells, it actually gives us a lot of information. That's what, as a computational person, we really love. So, for example, by looking at the variation between the different cells, you can ask about the dependencies between the different genes.

Wellington:  How did the different genes play into this? Are they working together and are they introducing more variation? 

Aviv: So, genes in cells don't behave in just a haphazard way in terms of which genes are expressed. They get expressed in these programs, you can think of it, or modules. Genes whose products are needed together are going to go up and down together. So now if you look at a lot of cells of the same kind and you look at how the genes are fluctuating or varying more subtly between them, these are not going to describe the genes that characterize the cell type. Because that you will see from looking at this cell type versus this other cell type. 

But they will characterize the programs that the cells are running. And so, you will be able to say, well, this kind of cell, it has these five programs. And in this particular individual cell, this program is higher than in that particular individual cell. And that might be physiologically very important. Maybe this cell is more active, say, a T cell that is more ready for the attack and this cell is more tired and more exhausted; it’s not going to give us as much of a benefit if you're thinking about immunotherapy in tumors. But it can also just give you this language of the programs that the cells are running on. And that's very valuable to us as we want to try and understand how cells act. 

And then we want to manipulate these cells and intervene in them and get the desired outcomes. Now you can take it one level higher. And instead of just thinking about one cell type in the tissue, you're actually looking at all of them at once. So in single cell genomics, you look at the many different kinds of cells and you look at many instances of each kind of cell. And you see all of them at once in one sample. And now you go to a second sample, and a third sample, and a fourth sample. And maybe you're looking at 70 different patient samples. Now, it's not just a fluctuation within one cell type of the genes, but you can see the fluctuations between the different cell types. 

So, in this individual, there might be a little bit more of these cells and there may be a little bit more in this state or running these programs. In this individual, it might be a little different, they have more of these. And as you look at them across many, many individuals, you're starting to see statistical patterns and you can say, oh, when I have more of these cells, I also have more of those cells. And maybe these cells are actually communicating with those cells and impacting how these cells are behaving. If everything is great you've just learned how healthy tissue behaves. If things are not good, you might find ways in which you can intervene in these communications. 

For example, you can find a receptor and when you have these cells expressing this receptor, you have those cells expressing this ligand, and intervening in the receptor- ligand interaction might be beneficial in order to shift the states of those cells. And then finally, when you go to the next level of what we call spatial genomics, where you get the cells organized in their native tissue setting, it's not just knowing about it from a statistical pattern point of view across samples, just counting cells and looking at their states. But you can even say, well, these actually like being physically close to each other because they're likely interacting with each other and doing something together. Now we can use this feature in order to understand the normal function of the tissue or the disease. 

Wellington: When you talk about the program of a cell, are you literally talking about the program of a cell in the way that we talk about the program in silica? 

Aviv:  Yeah. So, we actually think about it. I said I like thinking about cells. I'm a computational person, so I think about them computationally. So yes, you actually want to think about them a little bit in the same way that you think about code, that our genome encodes a lot of different programs. It doesn't use all of them all the time. But it's not just this aggregation of variables of the genes, but the genes are organized into these regulatory programs, first of all, which get controlled together through similar means. 

 So, they will respond to similar signals in similar ways. The outcome of that would be that, for example, when you need all of these components in order to perform some function, you were referring to pathways, so it could be a large protein complex. It could be a complex pathway. It could be a particular metabolic state of the cell. You will have the right components at the right levels. And in order to achieve this, these programs are actually encoded in the cell in several different ways. You have your genome that encodes, you know, your full capacity. But for a cell to actually execute on its genome, it needs to access that genome. It needs to access it in the right way. 

So, then you have a second layer, which is how the chromatin is organized to make some parts of the genome much more accessible to execution at any given moment in time than others. And that's really a process that unfolds on developmental timescales as we all start from one scale, from one cell and differentiate into different kinds of cells. And then you have the current state of the cell as it is encoded in the network of proteins and other molecules that you have inside the cell interacting with each other. They take the signals from the environment, they process them, and they process them in ways that we can actually model, in the same way that we model, you know, computational devices in the same premises that we often use in signal processing. 

You have low pass filters, for example. They get implemented by a bunch of proteins that have a certain response to a signal and produce a certain output as a result of that. And that gets kind of translucent inside the cell and gives you the output. 

What is difficult for us around cells, which is different than the way that engineers build, is that cells were not built by engineers, cells evolved through evolution. And we don't know a lot about them, right? We don't know all of the wiring and all of the components and all of the pieces. So, if you want to come to them with an engineer's rather than a scientist's mindset, we have to work in a world where 97 percent of the information sometimes is missing. You just actually can't model them in the usual way. You have to do a lot of inference in order to build up from data that you have to predict to first devise a model of what the cell might be. And so that's where it gets interesting from a computational perspective trying to figure out ways to decipher things where we don't know what they are up front. 

Wellington: Yeah. I'm struggling with that, because if in computational land, you start from the ground up. And if you say that 97 percent you don't know... 

Aviv: Yeah, well, that's an extreme case. 

Wellington:  …how does that analogy hold up? 

Aviv:  What is remarkably nice about biology and gives us an edge, even though so much about it is unknown upfront, is the fact that in biology, because of genetics, you have the ability to intervene in the causes. We actually know that if we do a genetic intervention and then something happens, if we devised our experiment correctly, then we figured out this is the cause and this is the effect. And so that takes us out of the place where everything is just correlations and not causation. 

In biology we have causation and it is directional. We have lab tools that allow us to do these interventions. If we do it in the lab and we have natural genetic variation, generated courtesy of evolution and the imprecision of biology, that lets us do it even in living humans because they're already genetically viable. We're not genetically intervening in them ourselves. And so now here comes the part that's so exciting for us today, is that today we can actually do experiments like these at really what used to be an unimaginable scale. 

We can go into a biological system and say intervene in every single gene at a time that we know about at least. And there's about 20,000 of them or so, give or take those that we're still discovering. So, you can go and perturb each one of them. And after you do that, you can come with these very rich ways of measuring or profiling biological cells. For example, with single cell genomics, you can profile their RNA or how their chromatin is organized. You can do it now by looking under a microscope and do what we call pooled optical screens and look at cell biological features under the microscope. You could do this and you can do a lot more and you can do it not just for 20,000 genes. You can start thinking about doing this for what we call higher order combinations, where you intervene at multiple genes at once and do what is becoming ridiculously big experiments. 

But the beauty of experiments like these is that they give you causation as a result of this. They give the right fodder to learn back the system, even though you didn't know what it was before it all. When you know a little bit, you can absolutely and should apply the knowledge that you already have, but you're not restricted to that knowledge. You have these more agnostic ways to get started and then you kind of bootstrap yourself for an answer. And the reason that this works for us is that in biology, we're trying to dissect. If you think about cells, you're trying to dissect the computational device. But the device itself answers your question, right? You go into the cell, you say, I want to understand you. I'm going to poke you, and you are going to deliver an answer to me through the phenotypes or the profiles that I'm going to collect out of you. And I think that's why biology is the most interesting place for anybody to work on, regardless of how they were trained. There's very few things like that that we can do in the world in order to understand the system where we know where the causes lie upfront. And all we have to do is do the right kind of poking in order to get the answer. 


Wellington: Okay, Stephanie, why did you start smiling when she said “poking”?

Stephanie: Ha, I guess it’s just my super nerdy sense of humor. I’m picturing someone walking up to a cell and going “poke”, “poke”, “poke”, “poke”, “give me the answers!” But honestly, that’s how it works in the lab. You poke or perturb a cell by changing its conditions. You can change the conditions inside - by introducing mutations or deleting a gene, or you can change the conditions outside the cell - turning up the temperature, for example. And then you watch how the cell reacts - or doesn’t - and there’s your data. You put all that together to infer how the cell is constructed. And that’s the way we’re going to reverse engineer the cell. 


Wellington:  So how is this changing the questions that we're asking? 

Aviv:  I think one type of question that it allows us to ask in a very ambitious way addresses the fact that biology is not linear. It's not additive. The whole is different than the sum of its parts. Meaning if I manipulate the biological system in one gene and I see what happens and I look at another gene and I see what happens, I can't actually predict just from those two results today what would be the impact of perturbing both. And so, until recently, our thinking was that if we ever want to know, “What is the impact of two perturbations?” or how two things relate to each other and so on, we actually have to do that experiment. 

And that was a really big problem. And the reason it was such a bummer is that it doesn't take a great expert to do the calculation of the number of possible combinations. It explodes exponentially. And it's not, you know, even just testing every pair of genes over a space of 20,000 is insurmountable experimentally. But if you think about three and four and five ways, right? Meaning perturbing five genes at a time over a space of 20,000, then forget about it ever. And if you think about human genetic variation, humans have a three billion base, three giga base genome. And then on that genome, they have about one percent genetic variation, common variation. I'm not talking about the rare variation. That number of combinations exceeds not just the number of humans on the planet now, it would likely exceed the number of humans on the planet ever and explodes quickly to the number of atoms in the universe. And then you're really screwed, right? Experimentally, it seems like you would never solve it. 

What is interesting now is trying to think about it from a perspective like those programs I talked about. So, imagine that you are now realizing, well, you don't actually have to test everything. You could test just subsets of things, and you don't even necessarily have to know which subsets are the interesting ones. You could sample the space of possibilities, and you're sampling far below the total possible number, because that's never going to be achievable, or at least for now, we think it's not going to be achievable. One has to be careful with that. But it's far above what we've ever done when we thought we had to actually test every possible hypothesis we're interested in, because then we used to be super careful and we would just direct to the few that we already had a really good reason to test. 

Instead, you were kind of in this middle ground when you're going to look at a lot, but not at everything. And when you analyze the data that you get back, you're starting to predict things that you actually never tested experimentally at all, because you're saying, “Well, if biology has this beautiful structure and genes go together in programs, then if I know what happens when I manipulate A, B and C and I get an outcome, and then I get a similar outcome when I'm manipulating A, B, and D, then maybe I can make a prediction about A, C and D, even though I never tested A, C and D and I never planned to do that.” 

And that's where things are shifting towards. And there's many examples of that, in how we think about patient data and how we think about lab experiments. And as a result of that, things that we thought were actually intractable might actually be kind of tractable after all. And that would be very cool. We will be a lot better at predicting what happens in biology. We will be better also at developing drugs as a result. 


Wellington:  Stephanie, is this what's called systems biology?

Stephanie:  Yeah, I think so. And good question, because it can be a little tricky to define systems biology. When Aviv said the whole is different from the sum of its parts, that's always been the key for me to wrap my head around what systems biology is. It's really different from what many of us trained in, which is a reductionist approach to biology. And the reductionist approach says, OK, this thing in front of me is super complicated. We're going to break it down and understand each part, one at a time. And in biology, that means focusing on one gene or one protein and studying it in exquisite detail. But what happens is when you put everything back together, it turns out this thing, this biological system, behaves differently from what you might expect. When you put all the parts back together, they change each other. They interact with each other. They feed back on each other. And that creates all this complexity. What Aviv is really excited about is we now have the tools to really start understanding all of that complexity.


Wellington: We're listening to you talk about programs and viruses, but it's also the biological viruses. You speak in two languages. Do you have two brains? 

Aviv:  So, I think a while back, I had two brains. I think they actually merged together and I like that. And I think it's also an important direction for the field as a whole to try and meld biology and computer science, or also with clinical and data chemistry perspective and so on. So rather than thinking about it as different pieces and the pieces talk to each other, that is the two brains metaphor. In the melding metaphor, the way I think about experiments right now, I think about them with a computational brain, but also simultaneously with very deep commitment to certain biological problems. 

The way I think about algorithms is entirely driven by biological motivations. And the biological ideas lead to thinking about algorithms in a different way than had I thought about them only from an algorithms perspective. I tried to look for the best algorithmic approach, to look at a biology perspective. So, I can't distinguish them anymore.And that's what I generally hope would happen. I think that the distinction is not helpful. Once you have the depth, you need the depth in each area and that you often need by separating the brains. But once you have it, you want to give them not just a very large interface, but to penetrate very deeply into each other. And I think that's where the world is heading. And that's something I'm incredibly excited about. 

Wellington:  Computers and biology have been with us actually a lot longer than people think. And perhaps, we need new language around this melding. I'm curious as to how these two brains have traditionally lived with each other and what you would say to people. You know, Jane used to talk about how she wished she could go back and get a degree because so much of the work is data-driven at a level you cannot do in notebooks. 

Aviv:  Yeah, so traditionally meaning for the past several decades in the interaction between computer science and biology, a lot of it can be described in the following way. And I'm doing it an injustice. By definition, it's impossible not to when you try to say that briefly, but it was often about taking a biological problem in data and encapsulating it into something that was well abstracted. Solving the abstraction in a computational problem. Applying the solution to biological data. Delivering the processed data back to a biologist for the biologists to have data that's better processed so that they can draw the insights. 

And so, I don't want to describe it this way because this is not actually how the community works. But you can think about it as two people playing tennis and there's a ball and they keep hitting the ball from one side to the other. I don't want to say a fence because it's actually very porous and you see each other and there's all sorts of exceptions to this metaphor where it breaks. But I think it's a pretty fair description. And it's not that anyone was the culprit of not doing it better. I actually think it was the best thing that could be done for a long period of time. This is where the tools were. This is where the data was: the volume of data that was in existence, the amount of biology we could measure, the type of algorithmic approaches that were at our disposal. 

Wellington: Was this the best approach? 

Aviv: It was the right solution for the right time. I think something is changing now and it's changing in several ways. Many of them happened really in the last maybe eight, nine years. The volume of biological data that we can collect is really qualitatively different. We grew in orders of magnitude, not in one, not in two, depending on the field in single cell genomics. It's the easiest to describe. Going from one to 10 cells or one to 20 cells to going to running an experiment with a million cells in a day is a pretty big difference. And it's still growing, but there's many others. 

Human genetics changed in the same way, dealing with the medical records, histopathology and digital pathology transformed this right across the board. Anywhere you look, it's like that, the way we do chemical screens today. And then at the same time in computer science, especially in the field of machine learning, there was a massive transition in the ability to work and learn from data without having models that were very bespoke. And that was driven by three simultaneous changes. 

The volume of data had to grow. Orders of magnitude. In order to allow learning like that, the ability to compute had to change dramatically. And that happened in this past decade. And then, of course, algorithmic shifts as well. So, all three together and you can see that in biology we're now delivering that scale of data. And so, if there's algorithms and compute, you can imagine that you would get the same benefit. And then the third and I think just as critical: our ability to intervene has changed dramatically in biology because our ability to do genetic manipulation is now just unbelievably clean and effective in large scale. And our ability to sequence human genomes, which gives us natural genetic variation, is also extremely scaled. So, when you put these three things together, now, it will be maybe a game with three tennis players, but you just can't solve it in this way. 

Wellington: How can you solve it then?

Aviv:  The algorithms need to change deeply in order to address interventional data. You can't just block algorithms that existing computer science modify them a little bit and they'll deliver for biology. The biological experiments need to be different and should be different because they could deliver so much more if they looked differently. And the way that we do human biology, which you mentioned in the very beginning, because we look at humans and human data, is so much richer and we have a real commitment to that. So that's why it needs to be more melded. 

Now, some of the melding is by having new people. People train very differently today. Jane said that you know, if she could only go back. But people who train today train often in two disciplines and in each of them deeply and rigorously without compromise. So that's one part. There's also plenty of people who just want to expand the domain. So, they're willing to say, well, I know this really well and I want to learn this new thing and I'm going to do it by first cohabiting with other people and working with them on problems together. And they're not being shy about saying, “I don't know this at all, but I'm willing to learn and I can.” I still can. I can learn new things. I am not beholden to what I knew before and for people opening up their domain to others to work in it. 

And in fact, in biology, this has happened multiple times in the past. Many of the pioneers of molecular biology did not train in biology at all. And that did not stop them from making molecular biology happen. And I think we shouldn't be shy about letting all sorts of people who didn't train in our respective fields and see that, in fact, they have tremendous talent and insight and inspiration and just go with it. 

Wellington:  What does that look like in 10 years or 25 years? What is the science? What does biology look like? 

Aviv:  So, let me think about this. So, I think several things. One is that the people who work in it, these distinctions that we make today, there is a computational biologist and there is an experimental biologist and there's bioinformatics and there's all sorts of other things. I don't think they'll exist anymore. It will just be a biologist. And because they study biological problems and they solve them. And the means will change in 25 years from now. Who can imagine even what those means will be? But the problems will remain. 

Biology is very big and very difficult as a problem. And so that's definitely something that would happen. I think also it would be a science with more prediction in it. I think today biology is not a very predictive science. We have to do experiments most of the time in order to see what would happen. It's hypothesis-driven science. There's a lot of hypothesis, but there aren't a lot of models that predict well in many realms of biology. And it is very humbling for that reason. It is tough and difficult, multidimensional and non-linear, things that are tough to predict. But I do think it will become more predictable and the predictive models will also be interpretable. Something that we don't yet have, even in computer science enough of. So that will play out for medicine. That would give us these kinds of roadmaps where you can imagine a patient coming in, getting worked up with measurements that we don't think about today at all as clinical measurements and having a model that relates to those measurements. 

And in letting the physicians make thoughtful decisions of, “This is this stratum of the disease. These are the cells. These are the processes that are manipulated. This is the pharmacopeia I have for this patient. This is how I will monitor them. I know that these resistances and other responses will arise over time or with high probability.”

It’s still going to be probabilistic. It's still going to be messy. It's biology, as I said, humbling. And there's evolution and it’s tough. But it's going to have more of that flavor. And if it does, then I think our generation will have something to be proud of. If it doesn't, we will have still tried our best. And that's all we can hope for.  

Wellington:  Aviv, thank you so much for being on the show!

Aviv: Thank you, it’s been a pleasure.


Wellington:  Well, that was cool, wasn't it, Stephanie?

Stephanie:  So cool.

Wellington:  I like how she imagines a convergence of all the disciplines of biology we've talked about.

Stephanie:  Yeah, I really love this idea that biology is actually interdisciplinary, that you can have different people, different training coming to the field and asking questions in different ways.

Wellington:  It makes me really excited to talk about all the things that have been happening in labs everywhere while everybody's been working remotely.

Stephanie:  Really true. I mean, things have paused or slowed down, but the work hasn't stopped.

Wellington: Let's get back to it then! Everyone, thanks for listening. If you've already subscribed, thanks. And you'll be the first to know when we launch Season Four, which is really soon.If you haven't subscribed, subscribe! So, you'll be the first to know! That's all we have for today. We'll see you soon for Season Four. And now for me, it's back to the studio!


The name Two Scientists Walk Into A Bar is under license and used with permission from the Fleet Science Center.