— Could you please tell us about your career in science? It's been quite a journey – from Moscow to the Weizmann Institute in Israel, then Harvard, and eventually MIT.
— I graduated from the Moscow Engineering Physics Institute (MEPI), then enrolled in a Master's program at the Weizmann Institute in Israel, and went there to study and write my Master’s thesis in chemistry. All my career decisions were motivated by my passion for science and desire to learn. I married early, when I was still in Moscow. My wife and I have traveled together ever since, moving from one country to another. Our first move was to Israel, where we both pursued our studies. Later, I got accepted into a PhD program in Biophysics at Harvard. We debated between staying in Israel and leaving, but eventually we both chose to continue our studies at Harvard — my wife pursued her interest in Psychology. Then we settled down in Boston. We've been living there for years. It was quite easy to move from Harvard to MIT – when you're already in Boston, not that big of a step. After completing my PhD, I stayed at Harvard for three more years as a Junior Fellow, it's an independent postdoc placement offered by the Harvard Society of Fellows.
This organization was established at Harvard in the 1930s to support scientists during the Great Depression. It has since evolved into this small exclusive club that admits eight people a year from different science fields. The fellows' only formal responsibility is to have formal dinners together once a week and have lunch twice a week for three years, that's it. Apart from that, they enjoy complete freedom of intellectual exploration. What's interesting about that place is that it brings together people from vastly diverse career backgrounds: composers, historians, anthropologists, and usually just one or two scientists — a biologist and/or a physicist. It's, like, this multi-purposed incubator with an extra rich nurturing environment. This enabled me to shift intellectually away from the tasks I had been working on in grad school toward the tasks that would occupy me later. When that placement ended, I started looking for a position and received several job offers, from employers in Boston and elsewhere. I chose MIT and have been working there since.

— So, your transition from Physics to Biology was your well-considered choice, not a spur-of-the-moment decision?
— Actually, there was no transition as such. I had wanted to do Biophysics from the very beginning, even in high school, that's why I enrolled in MEPI in Moscow to begin with. I had two excellent research advisors at the Department of Radiation Biophysics: Sergey Andreyev and the late David Spitkovsky. I wanted to do Biology from the start, but I wanted to combine it with Physics and computer modeling. To a certain degree, I was inspired by   Maxim Frank-Kamenetskii’s great book Unraveling Dna: The Most Important Molecule Of Life. Right around that time, computers started to truly appear in our lives. Some kind of computer center was built in our neighborhood in Kiyevsky District [in Moscow]. That's how I got the inspiration to pursue modeling and study mathematical models. No wonder, then, at MEPI I focused not just on Biophysics, but specifically on chromatin, the same topic that I'm currently working on: DNA folding.
Later, I veered off a little into a different area of biological physics. But some 15 or 20 years ago, there was this other big shift when genomics emerged as a separate and very promising field of study, and with it came new technologies and new data. And so I found myself gravitating toward genomics.

— DNA is your lab's research focus. How about we start off from there?
— Sure. Not only is DNA the most important molecule, it is also the longest molecule out there. It’s two meters long, packed into a five-micron cell nucleus, while DNA itself is only about two nanometers in diameter. It's hard to imagine, but let's try. If the nucleus were the size of a tennis ball, DNA would be about 20 kilometers long. Can you imagine how difficult it is to stuff 20 kilometers of thread into a tennis ball? It might be possible with an extra thin thread, but the question is, how does DNA survive this procedure and remain intact?
On the one hand, DNA is genetic material, and on the other hand, it is an array of physical molecules. It was also Schrödinger who highlighted in his book What is Life?: With Mind and Matter and Autobiographical Sketches that the genetic information carrier has to be a molecule. Later, it transpired that this molecule was DNA. But the thing is, the cell must be able to "read" the information. And so the real question is how these two roles combine: it is a physical object, like a book, and at the same time it is the text in that book. Access, reading, and packaging are the issues we are concerned with.
In many ways, we are driven by the new data that began to emerge in the late 2000s. Back then, a new method appeared that cast DNA in a completely different light than before. The traditional method is, of course, microscopy. You can observe the DNA inside the cell under a microscope, and you'll see this mass: individual DNA strands won't be visible, all you're going to see is this mishmash. Then it became clear that you could tag some places on the DNA and see where they were in the cell nucleus. This was discovered in the 1990s. And we could see that the DNA was not uniformly folded within the nucleus, nothing like that!
But the real game-changer was the arrival of this new technology that applied the genomic apparatus to obtain the ability to read DNA to understand how it's folded. This technology doesn't really tell you how the DNA is folded, but what it does tell you is which DNA piece is next to which other piece, or rather how frequently they occur together. In the end, I don't know “who sits where” in the nucleus, so to speak, but I do know which pieces frequently “talk” to each other, meaning how close they are. Now when this data started to surface, it naturally begged the question of what we should make of this dataset. Try to imagine what it looks like.   It's a map of neighbors in the genome.

— And this is where your unique method comes to the rescue?
— I did not invent this method. My colleague Job Dekker did. We've collaborated extensively with him since the very first article we wrote together. Here is how the modern biological method works. Sophisticated experiments are conducted, yielding a large number of DNA molecules that have been read. Lots of computation is needed to transform a basic list of DNA molecules into informative maps. That's where my lab comes in. Everything that transpires in the test tubes, and the subsequent feeding of these liquids into the DNA sequencer, is the job of the experimental lab. The machine generates DNA sequence files. That's our job, it's computational and theoretical work. Predictably, the first thing we do is, we simply process the data, converting it into charts that show where everything is located. It's the same as, for instance, taking Facebook data and visualizing it as a map of users who talk to each other frequently, as opposed to being just "friends". What we can see is who are friends with whom, but what we need is a map of each other's frequent commenters. This is an excellent analogy, by the way.
We figure out the users who are friends with each other, and we suspect that if they are friends as physical entities, located close to each other, then they can't be communicating over a great distance. Then we convert this information into maps. We have also developed a program where these volume maps can be viewed. It is similar to Google Maps actually.
When we completed the development of the program, we realized that the data we are dealing with really is the size of Google Maps, meaning that a single map derived from a single experiment creates a Google Map of the entire world with a resolution of 4 meters. But where we need to compare different types of cells, we will obtain 10 similar maps of one type, and 10 of another, and the result would be like a Google Earth map with a resolution of just a couple of meters, that's a huge amount of data. And it has its own continents, countries, cities – and you can even see the houses! The next question is, what do we make of this in terms of biology, and what do we make of it in terms of physics? How is it all folded? What are the processes that enable the folding?
This is where we need our computers, they really allow us to raise meaningful questions. Let's say DNA is all folded randomly. What would these maps look like if that were the case? We build a computer model of what DNA would look like if it were folded in the nucleus randomly or according to some rules. We generate the map we would expect to see, and compare it with the actual experiment result. In a sense, I would say it's done by trial and error. We try out different models to see which one works.

— So, you find the one that works and just discard the rest?
— That's right. You can think of it in terms of a Sherlock Holmes type of approach. We can't say exactly what happened, but we can eliminate a great many things that couldn't have happened. Proceeding in this manner, we select the possible scenarios and match them with our experiment. And we really hit on something astonishing there. Analyzing the maps and applying our computer models, we arrived at the conclusion that it is impossible to fold DNA in a cell in this way unless a special "motor" is used to fold the DNA in this highly nuanced manner. We voiced this hypothesis in 2015. We hypothesized the existence of a very special class of motors that do this thing we call Loop Extrusion, or loop formation, or loop growth.

— Is loop formation the pivotal aspect of the entire process?
— Looping has always been associated with DNA. Researchers were equipped to see DNA loops as early as the 70s. Loops per se are not the issue here. The key point is that there is a motor that uses energy, and it builds these loops. These motors are somewhat similar to the motors operating elsewhere in our cells, contracting our muscles, enabling us to move, allowing cells to crawl, and so on. The entirety of life within the cell and the whole body is powered ubiquitously by motors. This class of motors is well researched – they perform mechanical functions.
But the existence of motors that fold DNA came as a huge shock to everyone in the field. We had made a prediction based on theoretical reasoning, and it proved correct. Scientists came up with ideas about these motors at different points in the 80s and 90s — we found the proving articles. The question was asked roughly once a decade if DNA looping could possibly be the work of some motor. But those were merely hypotheses. What we did was come up with a theory. The difference is huge between a hypothesis and a theory. A hypothesis is a suggestion about how things may work, whereas a theory is a statement of how things really work, backed up by calculations.

— There was a detective story associated with this work, wasn't there?
— Indeed, there was somewhat of a detective story. We made no attempt to hide our work on   finding the motors. I had started telling people about it before the article came out, because I believed it was interesting and made a difference in our field. My presentations were received with a great deal of enthusiasm. We then posted our paper on bioRxiv. This is a widespread method of making your work public, especially in physics and mathematics: people post their preprint papers there before they undergo the journal review. Although, in 2015, this method of presenting one's work was new in biology. What happened is certain colleagues who had heard my presentations and seen our publication on the archive, had the audacity, to put it politely, to borrow our idea, our model, and add it to their already print-ready publication, making it look like it was their idea. Each acting on its own, two groups of scientists who had all attended my presentation simultaneously published their articles, claiming authorship on our theory. Many in our field knew what was going on. There are no effective tools to fight scientific piracy. It's just that, in our field, everyone knows who the real inventors are.

— Must be a good theory, huh?
— It really is! A bad theory isn't worth stealing!

— Could you tell us more about those motors?
— It would be easier for me to show them in action. Suppose I have a length of cable in my hands. Let's say it represents DNA. Then both my hands are a single motor capable of folding the cable into a tight loop. Every molecular motor is an enzyme that consumes energy and catalyzes a certain reaction. Some enzymes consume no energy, they simply catalyze reactions. But molecular motors do. There are other classes of motors: those that let our muscles contract, those that enable our heartbeat, and other types.
The gist of the hypothesis was that, first of all, DNA motors do exist and, second of all, there exist stop signals that will not let the motor pass a certain point on the DNA. We propose that, by the fact of their existence, the said motors generate a flow of traffic in the DNA. DNA traffic is the loop. Why is this fascinating? Firstly, the motors enable the creation of numerous loops, and thus a very long cable may end up being folded into a tightly packed system of loops. This is what happens in a cell that is about to divide. This is the moment when two meters of DNA must become four meters, and of those, two meters are to be packed into one nucleus, and the other two, into another. They have to be packed extra tight for that to happen. When we mention a chromosome, most people will picture a semblance of tightly compressed sausages. These only appear during the division of a cell. Well, the process of their emergence – and we demonstrated this in 2015 – is driven by the said motors.
Secondly, the looping process permits mutually distant DNA segments on the strand to find themselves close together and interact. For instance, the DNA segment that controls gene X finds itself next to the actual gene at the base of the loop, thereby activating the gene. In other words, the operation of these motors lets the DNA segments that control genes to exercise control. Therefore, the motors serve two purposes: One is to pack DNA for division, the other, to make it possible for DNA segments to control genes hundreds of thousands to millions of letters away along the genome. We used the “Sherlock Holmes method” to figure all this out. We theorists do not know how this actually happens. All we can say is that when the motors are at work, our observations (long, thick chromosomes, etc.) will be vindicated, but when other mechanisms are at work, the chromosomes won't be this way.

— Are there any alternative theories?
— Sure, there are. Moreover, when we claimed that a certain class of DNA-bound proteins are in fact motors, we were laughed at out loud. I was making a presentation once, and I heard someone in the audience shouting, "Nonsense! That's impossible! We've studied those molecules for 20 years. They aren't motors!" My colleagues who were there remember that moment well. Such heckling rarely happens at research presentations. I said, "How do you know they aren't motors?" They replied, "They don't consume energy". I countered, "Perhaps they can." — "No, they do not!" I said, "I think you've been studying a parked car. A parked car doesn't consume any energy and doesn't go anywhere. But that doesn't mean it can't consume energy or start moving". That was the actual analogy I made then. — "We don't know that".
Resistance was fierce, particularly among researchers focused specifically on these proteins. They had been under research for a long time and were thought to be merely hanging there, like shower curtain rings, holding two pieces of DNA together. We agree that sometimes they are indeed just rings. Like any car, which can serve as luggage storage while parked, but then the engine starts and it drives away. But molecular biologists that study DNA glad-handed our theory. It explains a multitude of things all at once, and that's often the hallmark of a successful theory — not a theory that explains a single phenomenon, but one that explains many of them simultaneously.
Incidentally, I got this definition of a good theory while having a conversation at the Harvard Society of Fellows with a colleague of mine, a historian, who currently works at the University of Oregon. I asked him, "What makes a good historical theory?" He said, "A good historical theory explains a host of different events occurring at a certain point in history, while also predicting the existence of a document that can be found, wherein we will find specific information". This person was researching the history of German newspapers in the early 20th century. Local newspapers were printed in villages, and people would share them with each other. It was an early version of the Internet in some sense. He was expounding the idea that this practice was in some way similar to sharing on social media, which would come a hundred years later.
It turned out that our theory had predictive power, because some motors won't budge unless fueled correctly (for DNA motors, ATP molecules are the fuel). And it explained lots of cellular biology phenomena, like chromosome condensation and the formative processes of specific domains inside the genome.

— What about experimental verification?
— When molecular biologists welcomed our theory in 2015, we knew we had to perform an experiment. But how exactly? For starters, we would try to break those motors or make the cell stop reproducing them, and see what happens. If our assumptions were correct, then some patterns on the maps would either disappear or change. Secondly, as I've mentioned before, there are these signals that stop the motors. It's not chaotic traffic like a rally in the desert. It's highly regulated traffic, with traffic lights, stop signals, and all the rest of it. We also had to disable the entire class of molecules that gave the stop signals, and see how the maps would change.
We did two experiments, followed by a series of articles in 2017 (we contributed to some of those papers, but not all of them). Our experiments confirmed that when the motors are removed, the exact things happen that we had predicted. If I’m not mistaken, five different articles came out in 2017 from five different groups that had eagerly joined the project all at the same time. All articles confirmed our theory.
But that wasn't the end of it. There exist numerous alternative theories out there that can be manipulated in such a way that they will appear to be informed by the same experiments. The leading alternative theory stated: "There are no such motors. They simply don't exist. That's not the way things work. Because we've never seen any evidence of those motors". Their main argument is this: "We've never seen a molecule binding to DNA and forming, or growing a loop that grows like this".
In 2015 and 2016, I consulted with some researchers who focus not on cells, but on individual molecules – it’s called single-molecule biophysics. They take a single molecule and examine how one single molecule interacts with another. It's an entirely fantastical field of study, in my view.
I used to work with researchers in this field, but at that time I failed to bring any of them over to my side. They all said, "Yeah, it's fascinating, but it's too complicated". This went on until a colleague of mine, a biophysicist at Delft University of Technology, Cees Dekker, who shares a last name with my colleague in Massachusetts, independently performed the following experiment. He took one of the molecules that we had predicted to be a motor, took a piece of DNA, pasted the material to the glass and inspected it under a microscope. And – lo and behold! – the molecule bound itself to the DNA! Cees provided a molecule with the source of energy, and it slowly began to grow a loop.
He sent me an email saying, "Leonid, we obtained an incredibly interesting result. Let's talk". We had a phone conversation. As it happened, I was on my way to a conference in Europe, but I made a stop in Delft first. We stared into the microscope together. Despite being a theoretician, I wanted to know how the assembly was set up. It was a really exciting moment. When I flew to Berlin from there, it felt like I was flying on my own wings. The whole flight I kept marveling at how spectacularly things had worked out. It was like a novel: a prediction, followed by experiments, a tsunami of skepticism and denial, then suddenly another physicist says, "We can do this", and he goes and does it! From that point on (early 2018) until just before the pandemic hit (end of 2019), a series of articles came out reporting similar experiments, demonstrating that other molecules of the same class behave likewise. There was this one crucial molecule that aroused the most skepticism. And it seems to be DNA's prime engine through all phases of the cell's life-cycle, except during cell division. Thе molecule is called cohesin. It stood there like a last-ditch outpost of resistance, and people were saying, "Well, those are the molecules that compress the chromosome... Because that's where compression is needed. It's obvious. Yeah, that's probably how it works. In fact, we've known all along that's how it works. But not with cohesin. It's impossible with cohesin because there's no compression involved, it's a much more subtle DNA traffic process".
I was at this conference in Austria in September 2019. All of a sudden, they announced a "surprise" presentation after the coffee break. Everyone began to wonder, "Could this be about cohesin being a motor too?" People were whispering to each other. Everyone returned to the auditorium. I was there, too, sitting in a side chair. The presentation was by the lab of a colleague of ours from Vienna. It wasn't him doing the presentation, but a postdoc of his. They had performed the same experiment that Cees Dekker had done earlier, except they used cohesin, demonstrating that cohesin is also a motor, and one of the same class. A collective gasp echoed through the audience. No applause, but this distinct audible reaction.

— A murmur?
— Yes, a murmur of sorts, the way it sounds in a small room seating about 150. That truly was the moment when everything was illuminated. The question of whether DNA motors exist was now resolved. Those motors exist in all life forms. They exist in bacteria, in archaea, and in all eukaryotes. They certainly exist in humans.

— What physical mysteries of biological molecules do you consider important to unravel in the foreseeable future?
— That's two questions at once. One is what those mysteries are, the other is what it takes to solve them. It's hard work to predict mysteries. What it takes to solve them is a good research question. There exist numerous gigantic international research groups focused on data collection. I do not question the benefits of data mining as data can be useful in many ways. But that's just not my style. A colleague once said to me (I loved the expression, and it kind of stuck in my memory), "My lab is an intellectual boutique". We love what we do. Every product, every article. Every article is one of its kind – we write it with lots of insight and with extreme care. Every sentence, literally every word we write together, as a team. We could say we run a boutique in that everything about it is creative, and every person is creative. We keep these huge balls of wool yarn in the lab that we play with to visualize DNA coiled up like a tennis ball. We paint pictures. My students ride bikes around the premises. They have fastened some bicycle tires to the wall. We have some old computer parts attached to the walls elsewhere in the lab. They took a glass pickle jar, stuffed it full of old memory chips, and labeled it "Memory Preserved". And so on. This creative vibe, this teamwork atmosphere I think is essential. It has to be there if you hope to unravel any "secrets".
Networking and artificial intelligence can help find solutions for lots of issues. However, in most cases neither of them leads to cutting-edge insights. Here, intuition leads the way, but intuition cannot be gained from either of those great things. And yet the data they generate is crucial. Data mining and machine learning are valuable. But a breakthrough usually follows the moment of intuition. The playful atmosphere we cultivate in the lab, and our playing with those molecules... I think it's the fertile ground of intuition.

— You're on this task force at the Massachusetts Institute of Technology researching carcinogenesis processes, and you've come with this unconventional idea about certain unusual mutations. Could you tell us?
— The story of how I got into this is interesting itself. At some point, the National Cancer Institute, one of the U.S. National Institutes of Health, figured it would be a great idea to get physicists involved in cancer research. I went for it. It sounded interesting, but I knew I had to do some homework first. I took a cancer biology course at MIT. I simply enrolled as a student, taking a semester-long course. It did me a lot of good. When you think about it, genomics provides a wealth of data about the specific mutations that occur in individual cancer patients.
Let's backtrack a bit. Cancer is an evolutionary process. It's important to keep this in mind: it's evolution occurring in the body. Individual cells mutate. If the mutation allows the cell to reproduce faster and thus be the first to become cancerous, then the mutation gets consolidated in the evolution. Herein lies the primary challenge of fighting cancer, I regret to say. No matter what we do, cancer will always find a way around it. It quickly learns to divide and, most importantly, to modify its genetic data. It all happens through random mutations. Random doesn't mean they may occur anywhere with the same probability, but we're definitely dealing with random mutations and selection. It's a Darwinian process alright. Then I figured I should take a closer look at the mechanism of the process. Does the cell know where a certain mutation needs to occur? No way!
Darwinian evolution: mutations occur everywhere, some are “good” and others are “bad”. So, how does cancer cope with the bad mutations? Could it be that cancer has no “bad” mutations? That would be too mysterious. We began to investigate this and discovered that apparently, any cancer evolves just like any other evolving population. “Good” mutations rarely occur there, but random and “bad” mutations abound (good or bad for the cancer cells, the reverse is true for the patient). Mutations that are good for cancer make it more aggressive. However, accrued “bad” mutations may suppress the cancer. Our perspective is that, in some sense, cancer represents an equilibrium between “good” and “bad” mutations. This led us to the idea that perhaps increasing the incidence of mutations could be a good strategy to fight cancer.
Any mutation is more likely to be “bad” than “good”. This might be what chemotherapy and radiation treatment do. They increase the mutation count, and since mutations are typically bad, they are effective. There are passenger mutations and driver mutations. Driver mutations enable the cancer to grow faster. All other mutations were termed "passengers" because they were thought to be neutral. But no one really studied them.

— They aren't neutral?
— The focus itself wasn't correct. In cancer genomics, it is typical to sum up one's work like this: we've collected data on thousands of patients, and we've identified these five mutations as being the most frequent for this type of cancer. Meanwhile, there are approximately 500 mutations in each patient. They have essentially discarded 495 of them because these mutations differ patient to patient. We dipped into those data waste baskets, and we found that the discarded mutations were not useless passengers at all. They were all “bad” for cancer. The cancer managed to progress despite those “bad” mutations, but it was a struggle. Now the question is, how do we arm those passengers so they can fight cancer more effectively? Or how do we load the cancer with enough passengers to halt its progress completely? That's the whole point. My colleague from Boston University, Michael Sherman, and I tested it experimentally. He felt quite skeptical about it. When you cultivate cancer and add mutations, traditional oncology says, "Cancer will grow faster", but we insist, "No, it will grow slower because random mutations will interfere". Despite his skepticism, my colleague and friend Professor Sherman started breeding mice with higher mutation rates. And it turned out, their cancers did not grow at all. They get cancer but it remains minuscule. It is unable to progress. We felt to be proven right: the so-called "passengers'' are not passengers at all.

— Was it some special breed of mice?
— Yes, the experiment was purpose-engineered to test our theory. Now we need to build on this and think how to translate our findings for clinical application. Another idea someone in our field came up with independently is that "passenger" mutations might be making the cancer more visible for the immune system. And so when we stimulate the immune system with one of the immunotherapy options currently in use, it is possible that cancers with more mutations will respond better to immunotherapy. One hypothesis that aligns with our idea is that "passengers" are bad because they trigger the immune system. This hypothesis is now widely accepted: the more "passengers", the better the cancer will respond to immunotherapy. We started looking into this data and, to my great surprise, we found out that this was not the case. The immune system is smart. It can see cancer after just a couple of mutations. The rest of the "passengers" are redundant. At this point, the cancer should have learned to fully suppress the immune system. This was another resonant paper, currently in the process of getting published. It caused a storm on Twitter when we posted it on BioRxiv. Many oncologists backed us, saying, "We know immunotherapy is effective for all patients". This was our key practical contention: there is no need to to indicate immunotherapy exclusively for patients with multiple mutations. Immunotherapy will work for all. If immunotherapy works for the given type of cancer, it should be given to all patients regardless of mutation numbers. We aim to continue in this direction, encouraging people to develop and apply immunotherapy more broadly, and to delve deeper into the mainsprings of the processes in question.

— Sounds like you've been involved in lots of collaborations. When experimentalists work together, how is the work organized? Is it a common cause or does everyone basically paddle one's own canoe, exhibiting degrees of friendliness toward neighbors?
— Good question. My work has always been largely theoretical. Collaboration opportunities involving experimentalists were hard to find, but something always came up. Now that my group has found itself at the forefront of research in chromosomes and physical configuration of the genome, we are receiving too many collaboration offers and are unable to accept them all. I hope my alumni, who are now setting up labs in different parts of the world, will be able to assist me somewhere down the road. There have been lots of new collaborations with people from many different fields where our ideas could make a difference. We have had good collaborations with immunologists, developmental biologists, and embryologists. This is logical, because, in a way, the role of the motors that fold DNA goes beyond DNA folding. I would say their primary role is different. You see, the genes account for only a fraction of DNA length, about 2%. As for the remaining 98%, we did not even know what that was. When we went ahead and tackled, with great gusto, the 98%, we found out that another fraction of a percent were genomic fragments, or DNA script that controls the genes and activates the right genes as needed. The amazing thing is how far removed the controlling segments are in the genomes of higher organisms. Suppose there is this chapter in a hefty genomic book that explains how to make pancakes. But the instructions as to when the time is right to make pancakes are in an entirely different chapter of the book, possibly even in a different part of the same volume, i.e. the same chromosome. But for one to impact the other (a fact discovered in the 80s), they must somehow touch each other. But how can they touch each other? Here is the thing. Apparently, the role of the motors we are talking about is to facilitate some manner of genomic scanning. One segment of the genome is "searching" for someone to talk to. "Friends on Facebook" is a good metaphor for how the genome is folded. As for who we talk to and who comments, that's an entirely different process. Let me put it this way: our FB friends are not necessarily people we often communicate with. As for communication within the genome... We're saying that our system of motors and stop signs is essentially a communication system that lets genomic regions interact over vast distances. Those regions can be centimeters of DNA away from each other. Yet they can get close, only nanometers away from each other, through the action of the motors. It is now becoming clear that this process may be vital to the immune system's health and development. There's a reason all cells and all life forms have those motors. It looks like communication between genomic components is entirely dependent on their operation. This is the most fascinating aspect at present. The success of this work hinges on the networking between researchers. The reason I am in France right now is I'm pursuing my new interests and I've built a new networking community with physicists and biologists who have a stake in these studies.

— We've been talking about physics serving biology this whole time, and I have a couple of related questions. Given this application of physics to biology, would it be appropriate to talk about a conceptual paradigm shift in molecular biology? Has physics itself benefited in any way from these biological studies?
— That's a great question, both of them are! To answer the first one, I hope it is a paradigm shift in our understanding of how the genome works. We all used to think the genome was nothing more than a one-dimensional object with some text. Then we found out that it matters how the genome is folded. Now it's beginning to occur to us that it isn't just passive text — the readers, as it were, of this text have a way of running across the text, connecting its different pieces, joining them together, and securing the correct action of the genes. Which is to say, there is an active, energy-intensive process (a motor in the physical sense) at work, steering the folding and activity of the genome. This is an entirely new concept – I see it now but I didn't when I first ventured into this field. Prior to that, I had occupied myself with protein folding and I thought the genome was just a really long molecule that probably folds according to the same physical principles. But I was wrong: it folds according to different principles where energy consumption and action of the motor are of the essence. Our biological studies have posed a new set of challenges for physics. From a physics perspective, some people on my team are polymer physicists, since our study focus is the folding of extra long molecules. I had developed an interest in this and pursued it back when I was still in Moscow, after reading the book Statistical Physics of Macromolecules by Alexander Grosberg and Alexei Khokhlov. Alexander Grosberg and I still talk occasionally. From a physics perspective, this field has raised a great many new questions. Big picture-wise, the way DNA is folded in the cell appears to dovetail with some hypotheses put forward in polymer physics during the 80s. But only now, 30 years later, new data has come in suggesting those hypotheses might be correct. New sets of questions have emerged that we are yet to find answers to. Here is perhaps the most basic question. You have a really long molecule, and there is this thing sitting inside it – not even a motor, just a small ring, but that ring can move around. How would this change the polymer's properties? Where would this mobile object be located? These questions are all new to physics. We are currently working on them.

— I watched your vibrant presentation on YouTube, titled The Importance of Not Being Serious. Why is it important to not be serious? Tell us about your hobbies.
— Well, I just wanted to do other things as well. I feel full of energy. When I was in high school in Moscow and I was a member of the Young Art Historians Club (at the Pushkin Museum of Fine Arts), I directed shows there and staged theater plays. I loved theater, I still do, and now I practice it as a hobby. We formed a Harvard KVN (Club of the Funny and Inventive [people]) Team at Harvard, then my friends and I formed the American KVN League. We toured the U.S. and the world. Our team went to Jurmala, this on top of the individual scientific pursuits of everyone on the team. It's really fun, but it's also a way to shift gears mentally. Moreover, I've always thought that art, in all its manifestations – even my amateurish drawings or KVN performances – is both fun and good for you. It gives your mind a jolt. You start to think differently and see things differently. I am fond of drawing, and I can see a great deal of similarity between drawing and scientific work, especially writing. When you draw, you can be strongly tempted to elaborate lots of minor details. This usually results in a dull and flat drawing. And then many details can't be drawn because I simply don't know how to draw them, so the drawing ends up being poorly done. The same thing can happen when you're writing a scientific article: as you write, you realize there are things you just don't know.
You try to figure out how those things might work, but it's not a cinch that you'll come up with something. You may and you may not. In a way, it would make sense to just be honest in your article and say: "Okay, we don't really know how this works. But we think it works more or less like this". It's like this bold, crude brush stroke versus fine detailing. You draw a line of a specific color, you intend it to signify a house, and our eyes can see that it does. But when we start detailing all the windows in the house, we end up with a tedious, unexciting image. The other factor is that we keep the reader in suspense just a little. We don't always reveal everything from the start. Similarly, a drawing only gets better when it has an unfinished quality, leaving the reader or the viewer some room for thought. I just went to this fantastic exhibition in Paris, organized jointly by the Musée Rodin and the Musée Picasso.
A dual exhibition like this, featuring Picasso and Rodin, is probably a once-in-a-lifetime experience. Rodin was a generation older than Picasso, and they never met. This exhibition tries to get the message across that, to an extent, both artists made similar discoveries as they blazed new trails in art. One of the things they both discovered (and there are references to Michelangelo in this exhibition) was the merit of the deliberate act of leaving things unfinished, infinito. They both practiced it, they intentionally left some works unfinished. They left some works incomplete, such as by leaving a rough stone (in Rodin's case) or leaving a drawing unfinished (in Picasso's case). A few days ago I saw Picasso's great painting Harlequin at the Pompidou. Only a part of the harlequin's dress is colored, and the rest is just cross-hatched — this is an example of intentionally leaving a work unfinished. I can see how this intentional incompleteness belongs in science, because there are things we don't know.
And where we do know, we might want to leave it to the reader to figure out, leaving it hanging. I also went to see Leonardo at the Louvre right before the pandemic. I had planned a stopover in Paris for a day between flights on purpose to go there. I noticed the same thing. Except I could not aver that Leonardo had left some paintings unfinished intentionally. St. Jerome, for example, is very obviously unfinished. I believe this concept of incompleteness carries a lot of weight in both science and art. We need to understand that there are things we don't know or areas where we can't intrude, and must leave them be. We shouldn't try to fill those gaps with arbitrary guesswork.

— I recall one other talk of yours where you spoke about kids' summer schools that you organize for your own and your friends' families. You teach "easy genomic science" there, to use your own words. Do you really think it's easy?
— Well, it's easier than physics, that's for sure! This isn't snobbery along the lines of the film Nine Days in One Year. Objectively speaking, physics is a vast and complex science. Genomics is more focused, and it's just a different kind of science. There's a lot of unknowns in it. But in this instance, I'm not teaching "easy genomics” because the science per se is easy. I'm simply trying to teach it to children of different ages through games and exercises. I teach a freshman course at MIT. We do similar problems with high schoolers in our camps. These are essentially exercises that help the kids understand the kinds of problems genomic science deals with. For instance, the kids may be asked to assemble genomes from short segments. We take a really long DNA strand, sliced into segments, and the kids have to put them together. It's a challenging algorithmic task. I could explain the algorithm and explain how to do it. But that would just be me, talking. There is no educational value in it. Instead, I take long printed strings filled with the letters A, T, G, and C, cut them up with scissors, pour the scraps on the table, and tell them to put them together for me. They can figure out that repetitions make no sense and are bulky, and they know that long strands are more valuable. If I had given them more copies, they would have done it sooner. I believe that awareness of one's own experience makes a big difference in teaching. They won't come up with a real algorithm to do the job, but they will capture the gist of the problem. I think it's more important to understand the problem than to find a solution.