Sergei Medvedev: When I was a child and a young man, I remember how the newspapers were constantly trumpeting: something is about to happen when the human genome is deciphered, when all these blocks and bricks become clear … And now the human genome has been deciphered - what next? A science called "bioinformatics" appears. What it is? Is the decoded human genome a kind of constructor, Lego, from which human life is created? Our guest is Mikhail Gelfand, bioinformatist, head of the Master's program Biotechnology at Skoltech, deputy director of the Institute for Information Transmission Problems of the Russian Academy of Sciences.
Mikhail Gelfand: I am also a professor at the Faculty of Computer Science at the Higher School of Economics and at the Faculty of Bioengineering and Bioinformatics at Moscow State University.
As I understand it, there are three billion letters in the genome. We know the code - what can we do with this code? This is some kind of cookbook of life, can we now cook a human, a homunculus from a test tube?
- This is the cookbook of life in the sense that life can reproduce itself according to the recipes contained in this book. We don't know how, we are bad cooks in this sense.
In general, this metaphor with deciphering and reading the genome is not very successful, because deciphering presupposes understanding, and so far we understand quite poorly. We learned to reproduce the DNA heredity molecule that was in a living cell, and then in a test tube, in a computer, we know in what order these letters are combined in this molecule. But understanding the meaning is a slightly different thing.
Bioinformatics appeared as an independent science exactly when biology gradually began to transform from a science that works with separate objects into a science in which there is a lot of data. At this moment, it becomes necessary to store, comprehend, analyze this data and do something with it.
This is about what years?
- In 1977, we developed methods for determining the sequence of DNA (I specifically say: not "decoding", but "determining the sequence"). Bioinformatics began to emerge, apparently, in the early 80s. I was terribly lucky: when I graduated from university in 1985, there was such a wonderful field in which there was no need to learn anything, it started from scratch, you could just take it and do it. This is very rare in history.
Promotional video:
Does it use more mathematical methods?
- The methods in it are mathematical in the following sense: you have to think. There in some places there are beautiful algorithms, beautiful statistics, but, in principle, the mathematics there is quite trivial, there are no mathematical magic wands there. You need a skill to keep in mind a lot of things and try to explain it in different ways, and the second skill is to ask simple questions. In this sense, mathematics education was very useful for me, not so much, perhaps, education, as communication with my grandfather Izrail Moiseevich Gelfand, who was a mathematician and worked a lot in experimental biology.
Now the genome is recorded, the sequence is determined - what can we do from this? I heard there is a new technology: we can take some gene chain and fix it, insert a good one instead. That is, we can operate with these letters?
- CRISPR is a genetic engineering technique, one of the very advanced, very modern technologies that allows you to do very precise and specific manipulations.
People just got more opportunities. In principle, people knew how to insert and remove genes before, it was just more difficult experimentally, not any manipulations were technically feasible. Now the set of tools has expanded. It was possible to build houses, as in Sparta, only with an ax, and now there is also a saw and even a jigsaw, you can cut out some beautiful platbands. In this sense, technological progress is very large, but so far not very meaningful. We understand some things: that there is a simple monogenic disease in which one single gene is broken - it is clear that if you fix it, you will have a normal embryo.
And this is already being treated?
- No, it is not cured, you cannot manipulate human embryos - it is simply prohibited by law.
But, as I understand it, it is moving. In England they allowed - with embryos up to 11 days …
- In China, they won't even ask anyone. You cannot slow down the ice rink by placing turtles under it: you feel sorry for the turtles, but the ice rink will be empty. In this sense, of course, it will move, but humanity needs to comprehend it. This is a really serious thing that needs reflection.
She's not the first. When genetic engineering had just begun in the mid-70s, when it became clear that genomes could be manipulated (then still bacterial), there was already a serious problem: for example, they were afraid that they would accidentally make some superbug and it would eat everyone. There were special conferences where rules were developed for what we do and what we don’t do. Any new set of tools expands opportunities, increases responsibility, and it must be understood.
Raises ethical questions …
- And if we talk about bioinformatics, returning to what you asked, then there is a slightly different story. There are two aspects to it. It turned out that we can answer quite a few classical biological questions simply in a computer.
I do a lot of genomics of bacteria. There are a lot of bacteria with which one experiment was made in their life, namely, the sequence of the genome was determined. We know quite a lot about them: what they eat, what they cannot eat, how they breathe, what they need to add to the environment, without which they cannot survive, but cannot do it themselves, and so on.
How much simpler is the bacterial genome compared to the human genome?
- It's not that critical. We share 30% of the genes with E. coli. In terms of the number of genes, a typical bacterium is thousands, and a person is 25 thousand.
Do you fully know which gene is responsible for what in bacteria?
- Not entirely, but we know a lot.
Much more than a person?
- As a percentage - of course.
The second thing that has appeared (and this, again, is related to technological development in experimental biology) and requires understanding in bioinformatics, is that we can look at a cell as a whole. A classic thing: a graduate student studies a certain protein, he knows the partners of this protein, he knows how this protein interacts with DNA, if it interacts with it, he knows when the gene of this protein is turned on and when turned off. This is such a full-fledged dissertation, several scientific articles about one protein. And then methods appear that allow you to answer the same questions for all proteins at once. For the first time, we have an integral picture of how the cell works; she is now very imperfect.
There is a protein that is unfamiliar to you, but you can predict by looking at its genome …
- These are two different questions. We are able to predict the functions of proteins without doing any experiments with them. This is a beautiful bioinformatics based on all kinds of evolutionary considerations.
Based on his gene profile?
- Protein is what is encoded in a gene, so it's better to talk about a gene: based on who this gene is with, who this protein looks like from at least a little known, on how it is regulated when it is turned on and turns off.
The same can probably be done about a person?
- It's harder. Technically, you can.
Look at the genome of some person at the embryonic level and say: a genius will grow or Down will grow
- This is a story about the fact that the function of a protein is generally unknown, nothing was known about it at all, and we can predict it. And what you are talking about is a well-known set of proteins, but with some variations - that's a slightly different story.
Man consists of known proteins
- Partly known, partly not. It turned out that we have a lot of heterogeneous information about how the cell works. The information is very imperfect, every single little fact can easily turn out to be wrong, but in the aggregate they are still correct. And from this one can try to describe the cell as a whole.
Molecular biology has long been scolded by philosophers for being a reductionist science. Here you are looking at the elephant in parts: someone studies the leg, someone - the tail, someone - the trunk, and no complete picture is added. Now it starts to take shape for the first time. One of the paradoxical results of this is that our knowledge and understanding in the absolute sense is increasing very rapidly. The progress in biology is amazing: we know much more than we knew 10 or 20 years ago, not even many times, but orders of magnitude more.
But the area of ignorance is growing even faster. That is, our relative knowledge is actually, in my opinion, decreasing, since it becomes clear that there are such open spaces about which ten years ago it simply did not occur to us that this could happen. And now we see that it is, but we do not know what to do with it. This is terribly cool.
It is clear who Down will be: an extra chromosome. But who will and who will not be a genius, we do not know how to predict, and thank God. We are not even good at predicting growth.
This information is not being accumulated?
- Of course.
Is it possible, for example, to compare a person's behavior, his profile on social networks with his genetic profile?
“I don’t know about that, but the psychological traits are partially determined by the genome, and they can be predicted a little.
Partly by the genome, partly by society
- Society, some life circumstances … In genetics, this is a developed thing, you can quantify the contribution of genetic factors to a particular trait. Let's take one - me. I have the same genomes in all cells, but my cells are different.
That is, at some point, genomes understand which cell to develop into?
- At some point, the cell realizes that it must become the precursor of the epithelium or the nervous system, or the liver, or something else. After the first divisions, all cells are the same, the genes in them work in the same way, and then they begin to work in different ways. The key thing is not really the genes themselves: I and chimpanzees have 50% of the proteins the same, and those that are different differ by one letter.
That is, the question is, where is the program that at some point tells the cell that it should develop into a person or a chimpanzee, and in a person - into the brain or liver
- It's in the same place, in the genes, but the key thing is not the genes themselves, but how they turn on and off. And this is the most interesting thing that is happening in biology now.
Is there a program that turns on and off?
- Sure. This is well known in fruit flies. Drosophila is simple, its embryo is also simple … No, Drosophila is complex, but the early stages of its development are very well described precisely quantitatively at the level of models. For example, you can predict the results of mutations. There are mutations when fruit flies grow a leg instead of an antennae. At the same time, it is known in which gene the mutation is broken, and this can be modeled - how the progenitor cells make mistakes.
Can it be fixed with new technologies?
- It is possible, but only in the embryo. When a leg or an extra pair of wings has grown, you can't fix it.
What can this bring in a practical sense? Let's say that what everyone is interested in is fighting cancer … With this amazing CRISPR technology, the Chinese seem to be trying to fight lung cancer. As I understand it, in this technology, a bacterium, when it sees a fragment of broken DNA, takes a piece from a healthy bacterium and replaces the broken chain with a healthy one
- Yes, just an interesting question, what happens to a healthy bacteria … No, not like that. CRISPR / Cas systems are bacterial immunity, a slightly different thing. When a virus infects a bacterium, if it does not have time to kill it, a war starts there, the virus switches some bacterial systems, breaks the bacterial genetic program and switches the bacterium to the production of new viruses. Actually, all viruses do this: bacterial, human, and whatever. There is a system that allows bacteria, if the virus did not have time to kill it at the very beginning, to cut out a piece of the virus's DNA and use it as a sample in the next attack of the same virus.
The bacterium inoculates itself with this virus
- In a sense, yes. And then it turned out that there is a protein that is able to cut a piece and purposefully insert it somewhere, and you can use this same enzyme for genetic engineering purposes.
I don't really understand about such cancer therapy: when you have billions of cells, how are you going to build the correct system into each of them? I don't understand how to technically do this. This can be done to treat genetic defects in the embryonic stage, when there is one cell.
With cancer, the story is a little different, there really is a very significant progress. It became clear that what we took for the same disease is actually at the molecular level - different diseases, and the targets for therapy should also be different. Cancer was initially classified simply by place: it was lung cancer, stomach cancer, skin cancer. Then histology began. When they began to look at the structure of the tumor, at what cells it consists of, diagnoses of the type of "small cell lung cancer" began. Then biochemistry began, they began to look at some markers, it shattered even further.
And now we can see what mutations actually occurred. You take a sample from a cancerous tumor and a sample from the same normal tissue, and see how they differ. They are very different, because with cancer everything breaks down, mistakes begin to accumulate very quickly. There are special terms - "drivers" and "passengers": some of these mistakes are passengers, they happened by chance, and some were drivers, they, in fact, led to rebirth.
There are completely practical things, because, for example, it is clear that some cancers, which were considered one disease, must be treated in different ways. Conversely, if you have externally different cancers, but they have the same molecular breakdown, then you can try to use a drug that is effective against one against the other.
Is it a breakdown at the genetic level, is some gene knocked out?
- Either knocked out, or, conversely, began to work too intensively. A typical sign of cancer is when genes that work in embryonic stages start working in adult tissues. These cells start dividing uncontrollably. Quite a few cancers are actually rebirth, degradation back in time.
I want to emphasize right away: I am not a physician, I know about this as a biologist and a person who reads a little reviews. I'm just always very afraid of disappointing people. There is always a balance between success in science and a practical issue - for those who will go for treatment tomorrow. These are experimental things. There is a single example where this worked. But it is clear that it is in this direction that everything will happen.
- If you look at the medical application, do you see that genetic engineering, gene therapy is already going on? Now, as far as I understand, individual autoimmune diseases show that one gene is broken.
- It is, rather, on the contrary, a defect in the immune system, a knocked out immune system. They are trying to treat it.
Immunodeficiency at the gene level?
- This is due to the specifics of the immune system. There cells divide all the time, new clones appear all the time. Even if you have everything defective, but you have made a small amount of repaired progenitor cells, they can replace the entire immune system, re-generate it. This is due precisely to the specifics of how the immune system works in general. In this sense, she is amazingly plastic.
Has the bacterium created some kind of vaccination, immunity?
- Yes, but it's a little different. Again, when it comes to immunodeficiency, this means that there are no classes of cells at all, because the gene that is supposed to work when these cells mature is broken. If you repair this gene to some precursors, they will mature into these cells, and they will give rise to this whole big immune picture.
There is also, as I understand it, computational evolutionary biology. Can you go back and see the gene for ancient humans?
- This is almost the most interesting. Bioinformatics is not a science in the same sense that electron microscopy is not - it is a set of techniques. The scientific part of bioinformatics is, firstly, what is associated with developmental biology, and secondly, it is molecular evolution, and there you can do various wonderful things.
We understand much better how it happened. Our differences from the mouse begin in the first stages of the embryo, and then everything is fixed. The same genes worked in slightly different combinations. This dream of describing the diversity of animals with an understanding of how they arose goes back to Haeckel. Haeckel juggled a lot, for which he is criticized, but the idea itself is very correct. To understand the difference between a person and a mouse, one must look not at an adult person and an adult mouse, but in embryos at the first stages. It is now becoming real.
The second thing: we understand who is related to whom, simply by comparing genomes. It is clear: the fewer the differences, the closer the relationship. This is a very simple idea, it can be algorithmized. Our ideas about the evolution of living things have changed quite a lot. Traditionally, mushrooms have always been studied at the department of lower plants, but in fact, mushrooms are not lower plants, but our closest relatives. Flowers with mushrooms are cousins to us. It follows from this that multicellularity has arisen many times independently, and this is already a very fundamental question. When you and I were in school, there were bacteria, then there were protozoa, and then the protozoa began to stick together and turned out to be multicellular, and then the multicellular ones were divided into plants and animals. There were some lower plants, mushrooms and higher plants - roses and buttercups. But in fact, not so:there were many different unicellular organisms, and in these different lines of unicellular organisms, multicellularity arose several times independently.
Man as the highest form of multicellularity?
- I do not know in what sense the highest. If you look at the variety of tissues, then all mammals are at the same price. If you look at the complexity of the nervous system, then we must be compared with octopuses. But if someone is pleased to be anthropocentric, then on health, I do not mind.
Our understanding of human origins has changed dramatically. In each of us, 2% are Neanderthal, and there were also Denisovans (Denisovans), about whom no one suspected at all. In fact, in Eurasia 40 thousand years ago there were three independent branches of humanity, they crossed in all combinations, and we see the remnants of these crossings in the genome.
Are you all taking over the remains of what was left in the parking lots?
“This is old DNA and analysis of modern DNA from different people. I think this is very cool. This greatly distorts my picture of the world.
Mikhail, you have puzzled us. 2% of Neanderthals, but there is a lot in common with mushrooms, with flowers … Indeed, here we are talking about the cubes from which life is arranged. Now, as I understand it, you combine these cubes in a different order, see what signs occurred in ontogeny and phylogeny, how the embryo of an individual person developed, how life on Earth generally developed
- Yes. We do it in a computer, and experimenters do it in cells.
We live in a delightful time! Let's hope that these experiments will lead to the creation of drugs for cancer and AIDS
- Actually, a cure for cancer has already been created.
I mean understanding the mechanisms of action
- And people diagnosed with AIDS live and live on modern medicines.
The question is not about drugs, but about how to treat it at the gene level. This is a further wish
Sergei Medvedev