MJM MedTalks
Genomes and Japan (Part 1): The surprising influence of viruses on human evolution

Introduction

00:00:04 Jan Pack (JP): Hello everyone, and thank you for joining us on another episode of the MJM Podcast. McGill Journal of Medicine, or MJM MedTalks, is a podcast series where members of the Faculty of Medicine and Health Science at different universities are interviewed on topics related to career, research, advocacy, and more. The aim of MedTalks is to open a space where faculty members can share information and advice for students in healthcare and medical science.

My name is Jan Pack, I'm a PhD candidate at the Department of Human Genetics at McGill, although I'm currently doing research at the Riken Institute in Japan, with which McGill has been conducting scientific collaborations. In this episode, as a member of the MJM podcast team, I will be interviewing Professor Guillaume Bourque.

Before I introduce our guest, let me first give you some quick background on today's topic, genomics.

The human genome consists of about 6.2 billion base pairs. Almost half of these consist of repeated sequences, also-called transposable elements, which have long been regarded as biologically inactive. However, we are now starting to realize that they may play a bigger role in the regulation of our genes than previously thought, and even give clues to our evolution.

Large parts of these regions are derived from endogenous retroviruses, or EVs, that are estimated to make up around 8% of the whole genome, making them about 8 times more abundant than our actual protein-coding sequences. EVs are genetic sequences, including long-terminal repeats, or LTRs, that were likely inserted into a genome by a viral infection during our evolutionary history, and have since influenced the way our genome regulates itself, as well as being involved in the occurrence of a range of diseases, including cancer.

Today, I'm talking to Guillaume Bourque, professor at the Department of Human Genetics at McGill, and scientific director at the Canadian Center for Computational Genomics. His goal is to understand mammalian genomes by comparative analyses and investigate the role of transposable elements in gene regulation and the impact of genome rearrangements in evolution and cancer.

He has also played a major part in the collaboration of McGill University with Kyoto University in Japan, one of Asia's strongest centers of biomedical research, which has been running a highly successful joint PhD program for genomic medicine in the last few years. This collaboration recently culminated in a series of publications on genomic research that gathered much attention from the press, such as CNN Science. First, we will talk about his current research and then discuss his experience with the Japanese-Canadian cooperation.

Interview

00:02:55 JP: So to start, how about you give us a quick introduction in colloquial terms about your own research?

00:03:04 Dr. Guillaume Bourque (GB): Well, first of all, thank you for the invitation and the opportunity to discuss science with you today. So, I mean, I think you did a great job of summarizing some of my interests. I was really amazed by the fact that so much of the human genome and how it works is still unknown. And I'm quite passionate about sort of this dark part of the genome that initially was described as a repetitive element.

It's also known as these regions of the genome that, as you mentioned, have this ability, well, some of it come, you know, is derived from viruses, which infected the genome many million years ago and had the ability to create copies of itself within the genome.

And initially, people really thought that this was indeed just junk DNA, some kind of a weird process that's happening and just is tolerated by the human genome, but doesn't really have a consequence. But more and more people now realize that actually it's a way the genome has been evolving. And so, yeah, I just find that quite fascinating.

And so I started my lab in this space and then since then, continuing to explore these questions.

00:04:35 JP: Yeah, that's great. I totally understand. And I think we like to think of our own evolution as being kind of purely ourselves, whereas now we're kind of realizing that there's also other factors from outside that might play a role in that, which is super interesting and now kind of getting behind the idea that those huge chunks of our own genome that we thought were just inactive and doing nothing, that they might actually play a much bigger role than what we thought.

So when exactly did you establish your lab and what drove you to do that? What was your motivation behind that?

00:05:11 GB: So actually, well, I'm currently in Japan, you know, because of this joint PhD program and so I go back and forward between McGill and Kyoto University but I actually started my lab in Singapore.

So I visited Singapore as part of a conference and met some people, ended up starting my lab there all the way back in 2004, and so it was quite a number of years ago, over 20 years ago. And that's where actually I started to work on transposable element.

I'd never worked on transposable element before that, but I was working at that time at the Genome Institute of Singapore, and they were developing some of the first functional genomic assays and that allowed to look genome-wide at what was happening, what were the binding sites of P53 and some key transcription factors.

And it was the first time that people could look at that genome-wide, and that's where we observed that a lot of the binding sites for P53 were not just in promoters or near promoters. I think 30% were in specific families of ERVs. And then we were like, whoa, what is going on? And so that's how I actually started - I mean, I guess I had already started my lab then, but that became after that the focus of my lab. And I ended up spending six years in Singapore, before moving back at McGill in 2010.

00:06:50 JP: Yeah, so you've been around!

00:06:52 GB: I guess so!

00:06:54 JP: Must have been an exciting time to be there at the time where we kind of started getting this more bigger picture technologies where we can realize how much more is going on.

00:07:03 GB: Well, it's funny you say that because when, and especially given that this is a career podcast, I mean, it's strange how different periods have different types of excitement.

I mean, right before I moved to Singapore, the excitement was the sequencing of the human genome in 2001 and that was the big, big buzz, it was the fact that the human genome had just been sequenced. And at that time, it also felt a little bit like, well, what else are we going to do? Because now we've sequenced the genome and then we're done.

But indeed, as you're correctly pointing out, then there was this phase of like, oh, now we have the first time some of the tools that allow us to study the genome, and we still don't understand how it works and so at that time, it was indeed very exciting because this was the, you know, there was this big change with the new technologies of short read sequencing that became, you know, this is when Cellecta technology became Illumina technology, and all of a sudden, we could sequence and do all these things. So indeed, there was a lot of excitement.

But I would argue that it's exciting now as well. I mean, right now I'm involved in some of these new projects around pangenome and the fact that we can actually sequence, not just one human genome, but many human genome that had the same level of quality as the first human genome.

But now we can use this information that we're not identical and really including studying these repetitive elements and difficult parts of the genome and how they differ between individual. And, you know, of course, we've been studying variation between individual, but mostly at the level of SNPs (Single-nucleotide polymorphisms) and simple regions of the genome but our genome changes in some of the other regions and I could not believe how much human DNA is not represented in the current human reference genome, because every individual, you, me, we have megabase of DNA that's not in the reference (genome), right?

So what does it do? Is it not important? I mean, to me, that's pretty exciting as well. And then, again - so I think different periods, different tools and different questions, and there's potential all the time, I think, to be excited about. But it's like- I mean, for me, it's really a question of then, not being afraid of jumping on the new technologies and applying them to new questions that come around.

00:09:47 JP: Being open to new methodologies.

00:09:51 GB: Yeah.

00:09:51 JP: At this point, you mentioned something that might be kind of good to really quickly explain, which is the reference genome.

00:09:58 GB: Yes.

00:09:58 JP: Having that and what that means for this kind of research has kind of really big implications. So, if you could really quickly summarize what that is and why it matters.

00:10:08 GB: Yeah, well, so you mentioned in your intro, the human genome, it's 6.2 billion base pairs but typically, we think of that, and then we think of that as a fixed concept and we think and we have a reference, the first genome that I mentioned that was sequenced, you know, it's one representative, and it's not even the genome of one person, it's some kind of a composite, but we're using it as this represents the human genome, and it's one version but, you know, your genome, my genome, we don't have just one copy of the genome right there. The reason it's 6.2 is because there's the maternal haplotypes, the paternal haploypes. So that the reference is only one version of this and it's only 3 billion base pairs.

But now that sequencing is getting easier, we can relatively easily replicate the work that was done 20 years ago, but now sequence your genome, sequence my genome. But the question is, how do we use this information? Because when we do an experiment, typically we still then rely on the reference human genome to map all of the reads that we get and all of that.

But maybe it would be better if we did an experiment using your cells to then use your genome to study how your genome is acting in your cell, right? And we just didn't have the technology and the tools to do this in the past, but now we're starting to have that. And so we definitely have the technology to sequence your genome.

What we're still working on now is the technology to use your genome and all genomes that have been sequenced to build a new reference that's going to contain, maybe not all of the genomes, but all of the representative genomes from a very diverse population.

Like, okay, let me take an example of, since we're going to talk about Japan and McGill. To me, it's a little bit crazy that if you have a Japanese project, that's characterizing a Japanese cohort, that you take all of this data and you map it into a human reference that is probably of European descent, right? So why not use a reference that would represent the Japanese population and then use that as a starting point?

This is the work that's now happening, and we're calling this the Human Pan Genome Project. It's to create a reference that would be representative of genetic diversity and would contain the genome of individual of different ancestry and would hopefully then ensure that some of the downstream genomic and genetic analyses that we do, you know, would use that as a better starting point.

So, again, I think, you know, this is also quite worthwhile and ambitious as a project. So, this is really like human genome 2.0, right? So now we want to have a human genome reference that is richer and represents genetic diversity. And how do we do that? I mean, we're still working through the details of that. So it requires changing a lot of the algorithms that we use for genomic analysis. It requires quite a lot of work.

00:13:39 JP: It's definitely worth the effort… Yeah, that's a great point, and it's important to remember that we might be talking about only two or 3% differences between the genomes, or it can be more, it can be less, but in terms of DNA, even just a few base pairs can make a huge difference, so..

00:13:58 GB: Yeah, that's a great, great point. So to me, the two spectrums are very interesting. On the one hand, we want to study what is shared and common between everybody and what distinguishes humans from other species and things like that.

I mean, that's, I think, very interesting and important. But as you just said, we also want to study what is different between individuals. I think especially in the context of thinking about medicine and how different people react differently to drugs or are susceptible to disease in different ways, this comes from only a small portion of our genome, but that portion is quite important to understand such that we can hopefully adapt and do a better job in terms of treating people or preventing diseases, right?

00:14:53 JP: Yeah, definitely, and I think that plays quite well into the next question, which is that we would like to think of our genome as something maybe static or something that only changes maybe during sexual recombination or maybe through mutations but having things like transposable elements that move through your DNA, it might occur much more frequently than what we usually think of.

So how, if we think of our own bodies, how can we imagine the genome changing in each of the cells as we go through our life? Is it- will they look vastly different to each other by the end of our life, maybe say, when we are 80 years old, or will there be almost no difference? It's just one or two events every couple of years? How can we imagine this change occurring?

00:15:44 GB: Great question. I mean, we don't actually know the answer to this question yet.

Because the tools that we use to try to characterize how much the genome changes, it requires us to use many cells. So, we have to pool samples, like even from one individual, but it's very difficult to profile the genome of one individual cell. So a lot of our estimates are quite crude because we have to take many cells together and then try to detect and observe changes.

There's still some contentious studies, you know, trying to estimate these rates of change and in the US right now, I'm aware of one large consortium that tries to systematically answer this question and is profiling using some of these new technologies, the amount of change in different cells because the amount of change might differ in different tissues, it might differ based on age, based on external condition. So there's still lots of questions there. But I think it's clear, though, that there are some changes. So we do know that.

What we don't know is, again, the rates of change and whether and what's affecting these changes. I mean, it's also clear that it's relatively stable. So it's not like we're changing that much. But just like we just were saying about genetic variation, you know, even if it's just one or two percent, but if it's associated with something that's very important-

So like one area that people think that somatic change might be happening is in brain development. So they think that, you know, in brain development, you have, of course, a lot going on in terms of the way sort of the networks of cells and how they're connected and all and how that develops during development and later in life. But there's quite a bit of evidence that there are also genetic changes that are happening in- during development, during brain development as well, potentially physically changing the brain cells and the neurons.

But, again, estimating those rates and even harder is then trying to figure out if it is really some cells are different, what does that do and how does that affect the brain? I mean, these are very difficult questions because, again, if one cell is unique and different from all the other ones, how do you figure out what exactly that specific cell is doing? You know, it’s really quite difficult because, again, most of our assays work in bulk so you need to repeat the experiment and things like that. But if all of this is really happening at the level of a single cell, it's really quite challenging to understand how it works.

So, I mean, this is an area with definitely some work currently happening that's also quite challenging and you know you could combine that with the previous topic that we talked about with the fact that there's some genetic variability between individuals and then ask* how does the genetic variability combine with the somatic variation to explain some of the differences that we observe between individual when, again, like we're, I think we're far from being able to answer these questions to some extent, right?

00:19:45 JP: Yeah, big questions, definitely. It's crazy to think about that we might be changing a lot more than we would think.

00:19:52 GB: Well, to be honest, I think, you know, if I had to gamble and say, where do I think most of the exciting biology is, I actually don't think that it's in that area personally.

00:20:08 JP: Where would it be instead?

00:20:10 GB: Well, because I think, I mean, like thinking about transposable elements, one thing that I've seen is that the environment affects our cell epigenetically and how they're organized by, an order of magnitude more than genetics, right?

So, yes, genetics are, and I'm a professor in genetics, I have to, you know- so I think genetics is very interesting because it explains risk for certain diseases. But if you think about how it affects risk of certain disease, it's nothing compared to some of these other environmental factors. You know, if you're smoking or if you're exercising and all of these things actually affect your cells in ways that are much more drastic than some of the more subtle genetic differences between individuals, which is, yeah.

So anyway- so I kind of- I like genetics, but I also like epigenetics for that reason. And if you think about epigenetics, you know, and how, again, like most of our genome is almost the same, but obviously brain cells and blood cells are completely different, right? And they have the same genome.

And so they might have some tiny little genetic differences, but epigenetically, they're crazy different, right? And this is in development and then in, the environment affects that., but, I think the epigenetic changes are quite dramatic as well and very interesting to study in this context.

00:22:05 JP: Let's get a little bit more in depth into one of the recent publications, just to kind of highlight a little bit why this is important to look at and how it might have affected our developmental histories. Because in the research paper that came out with this collaboration between McGill and Kyoto University is that you looked at the impact of transcription factor binding.

Most notably the SOX family, S-O-X, which can bind to the repeating sequences of endogenous retroviruses. And we know that the SOX family is an ancient group of transcription factors that is known to play a very important part in the process, such as sex determination and neurologic development. So yeah, if you could, freely speaking, what kind of impact would you imagine these kinds of changes could have on the human development itself?

00:23:02 GB: Thanks for highlighting that work. So, I mean, this goes back to one of the first things I said, like the fact that we still don't quite understand the human genome and we talked about the fact that 50% of the human genome is repetitive and is derived from repeats and eight of that 8% comes from viruses.

But when we say all of these things, even those numbers and those statements, you know, we're still refining them and understanding how, and- So we don't really know what percent of our genome comes from viruses because there's more recent viruses in our genome. We can recognize them as viruses, but some older elements of the genome, we might not be able to tell that they're viruses.

So, we're still trying to characterize what's in our genome and so that particular paper is revisiting a little bit, this 8% that comes from viruses and in that work, we made an attempt to basically revisit the annotation and I tried to see whether we had the correct annotation of which virus families were there.

The way we were doing that is that we use it a bit like what we were saying, some of the new technologies to test and, the sort of, to generate some functional data on these elements and we use that to propose a new annotation of these viruses, which we think is more accurate compared to what was previously done, which, again, dates back to when they sequenced the human genomes. It's not so surprising that we can improve a little bit in the annotation using some of these new technologies.

We did all of that, and then we also wanted to show why it's important to do that. And why it's important to do that is because with the new annotation, we could then revisit exactly when did a specific group of ERVs in our genome spread in the genome and we were focusing on one specific subfamily, so one virus infected the ancestor of human and chimpanzee based on our annotation, which again is just a refinement compared to what was there.

But using that, we could say, OK: this family of virus infected the ancestor of human and chimpanzee. And then we can look more specifically: how did it differ compared to previous viruses that infected older primate family?

And we could observe, as you mentioned, this SOX motif that really distinguished this subfamily that had infected the human chimp ancestor relative to macaque, and then suggested that, maybe the fact that there was a binding site for that transcription factor maybe contributed to that specific family spreading in that ancestral primate genome.

Exactly what it's doing and why? I mean, there's still some question, but we use that as an example to show that with better annotation, we can also then make these observations about what happened in the evolution of the human genome. And yeah.

00:26:40 JP: Very nice. It's a very interesting paper, I encourage the listeners to check it out. I'm going to link it in the description box.

00:26:48 GB: Yeah, Thank you.

00:26:52 JP: Yeah, and kind of connecting to that in the genomics field, especially when we talk about interactions of transposable elements, we also have our own elements that try to repress those in order to protect us and this kind of, let's say, fighting or co-evolution is sometimes referred to as the arms race.

And usually what we understand as an arms race, in modern history especially, those always have an unexpected side effect that drive technological development in new ways that we didn't expect before. So could we see any of those unexpected advances during our evolutionary history as a result of this genomics arm stress?

00:27:38 GB: Well, great question. I mean, maybe by the end of this podcast, I'll convince you to study transposable elements!

00:27:47 JP: We'll see.

00:27:48 GB: We'll see. But no, it's a great question. So absolutely. So I think this is why they call it an arms race between. So if we take the simple, well, I say simple because the viruses is one type of transposable element, but let's focus on that type, because I think that's the one where this arms race is the most obvious because you really have the viruses that are competing in a way with the host, the human in this case, the virus are trying to use us to spread more viruses. We're trying to control it.

Indeed, I mean, this sounds like a war, right? Which is not a very, you know, it leads to destruction, not to creation of innovation and so on. But turns out that, these viruses that are integrating the genome, the ones that are successful and the ones that we, you know, the ones we were talking about that basically survive in the genome and make it, many of them end up indeed creating innovation in our genome and helping us in all sorts of things.

I mean, they've contributed to placenta development. They've contributed to immunity, which is pretty strange. So some of our immune defense comes from viruses, which is very strange. So we've sort of integrated pieces of viruses to, you know, train us to fight other viruses, which is pretty, pretty strange. So absolutely. So it leads to a lot of interesting-

Well, I'll give you just one more example, which again goes back to how much the human genome is full of things that we don't quite understand. So in the human genome, one of the largest class of genes, and especially of transcription factors, are zinc finger protein transcription factors that have these zinc finger domains.

And if you look at the gene list from the human genome, especially the transcription factor, I think, I don't know if it's a third or like in the largest family, there's like olfactories and receptors, there's a large family of zinc finger, and they're all named, you know, ZNF100 and ZNF101 and so on. All of them are sort of, they didn't know what they were doing for the longest time because most genes, again, don't have and are not known exactly what they do.

There was this amazing observation that the number of zinc finger protein in organisms was highly correlated with the number of transposable element families. So people realized that they look like they have -- and initially, it was just like this observation -- that it looked like the more you have transposable element families, the more you have these families.

And finally, people realized that one of the main function of these proteins is that they're contributing to the arms race and they're, you know, often are used as a host. It's a gene family that expands. So you use when a new virus attacks, if we take viruses, a new virus comes in and starts being active in the genome.

Often the host duplicates one of these one of these transcription factors that was basically repressing a related element and tinkers with it, changes it a little bit, and all of a sudden, then the host has a defense mechanism against these new families.

You could think that this is only in the context of defense, but it turns out that it's used in this defense arms race component, but sometimes then the host also uses - the host being us - then says, I just got this new transcription factor for free it's doing this, but maybe I'll also use it to do something else because it turns out that this arms race leads to a lot of innovation in terms of regulation, but also in terms of new genes in the human genome, including all of these new genes, some of which, again, we know what they're doing. Their main function is repression of these TEs.

But more and more, you know, people realize that they're doing other interesting things. And one of the reasons why they're not so well studied is because they're all relatively new. And to go back to one other thing we were talking about, a lot of people study things that are conserved between human and mouse, but these are examples of families that are actually, these are genes that are quite new in the primate lineage because they serve this function associated with transposable element that are new in the primate lineage. So, you know, it's a lot of new biology and people have just not been studying it. So very interesting question and area.

00:33:13 JP: Yeah, very interesting response, thank you so much. It's fascinating to see this interplay between the humans and the viruses. If I got infected with the virus today, I wouldn't have expected that it might help me, you know, a couple of million years down the line. So-

00:33:34 GB: Well, that's right. I mean, obviously only a subset of the infection lead to these good outcomes, right? I mean, one thing that's crazy about all of this, of course, is that none of the infection- like to go back to this idea of the genome changing, right? None of the infection, you know, really affects the next generation. Infection leads to genetic change in your somatic (body) cells in some cases, but for this to really have an impact long term, it needs to affect the germline and so, all of that is obviously quite rare because as a species, we wouldn't want that. So most infections, you know, don't affect what's going to happen in this context.

00:34:22 JP: Which is another question of looking at how those sequences get into the germ in the first place, since it is such a protected place, right? So that's another question. I'm sure we could talk a lot about that. But going a bit more back into the big picture, also something that we mentioned already earlier is understanding the genome better also influences how we understand human health and our ability to act on that knowledge.

Could you comment a bit on how understanding our cells, our genomes, maybe even our epigenomes better, can help us down the line to prevent diseases and treat diseases and understand diseases?

00:35:07 GB: I'll go like sort of, I mean, we could go very deep into this, but one key component of this, of an answer to your question to me is I think we're now very good at sequencing genome and identifying genetic variants. We can take one individual that has a rare disease or a cohort of individuals affected by a common disease and really look at genetic variants that distinguish these individuals, right?

What's very difficult is to interpret those genetic variants, know which ones are really important and which ones really are contributing to the disease. So a lot of the things that I talked about is really about understanding the human genome, such that when we observe genetic variants in the disease context, we can interpret them better, prioritize and identify really what genetic variant is important for that disease because a lot of times we might have, you know, we sequence this individual that has this rare disease.

We detect maybe 100 mutations in this individual that we've never seen before. Which one is the one causing the disease? That's still difficult question to answer because we don't know what all of these hundreds of mutations or variants are doing.

But as we develop better understanding of TEs (Transposable Elements) and zinc finger proteins and then just the human genome in general, we can start prioritizing and understanding which of the hundred variants are important. And then from there, do something which is also then very difficult, which is maybe change, you know, how we would treat this patient.

But at the very first step, at least having a better understanding -I think that's the kind of problem that I'm hoping to help with some of the work in my lab - is really help assign function to variants that we would detect in the individual that had a different disease.

One big challenge too that I've been thinking about is that everything is sort of interconnected, right? So a lot of times we oversimplify things and we think about this variant is doing that. But in practice, as we were saying, the same genome is found in all of your cells. And so, you know, nothing is acting alone. And so you pull here, it's going to pull there.

And I think we're also very far at understanding sort of all of the downstream effects of trying to make, you know we might be getting closer at using CRISPR and correcting mutations, but do we really know what changing something really does, because it might do that in this cell type, in this context for this disease, but we don't necessarily know what it does in another cell type for something else, right? And so that's there, there's still lots to be done.

00:38:36 JP: Yeah, absolutely. We keep discovering more things and how those things affect each other, and I think it's good to remember that just before we managed to sequence the first human genome in the 2000s with the Human Genome Project, the general sentiment was that “Oh, once we have the sequence, then cancer will be a thing of the past”, which is, of course, not what happened, right? Because that was only a starting point of finding out so much more that influenced more things that we had to find out.

00:39:09 GB: That's right. So I think to me, it's clear that we're making a lot of progress, but it's also clear we continue discovering layers of complexity, right? And layers of things that are important that we didn't even know even existed, like some aspects of this genetic diversity between cells. And then there's just much more complexity than what people anticipated and so still lots to be discovered and I'm sure there's even more, right? I think we're still oversimplifying the complexity.

00:39:49 JP: We don't know what we don't know, right?

00:39:51 GB: Exactly.

00:39:53 JP: Yeah. And kind of connecting to that is also what you mentioned earlier, that there's always some excitement and some new technology that is helping us making breakthroughs. I think for our current generation right now, or since the last two, three years, is artificial intelligence that's kind of booming of course.

So we know that has influenced a lot of life science sectors already, for example, with AlphaFold that can predict protein structures instantly almost, or AI-assisted drug development. So let's say in the coming 5 to 10 years, how do you think that machine learning or artificial intelligence will influence genomic research and change our views?

00:40:44 GB: Well, great question and great follow-up to the previous thing we were just discussing about the complexity of some of human biology. So, I mean, number one, these new technologies are already changing the way genomic research is being done. I think with great and also mixed success, you mentioned some really exciting development with things like AlphaFold.

And so I think definitely there were some areas in genomics that were really primed for this type of technology where you already- there was already very, vast data set that could be mined by these kinds of approach to really improve prediction. And people have tried over the years all sorts of methods, and all of a sudden now, we have more mature methods that really outperform everything we'd seen before. So on one hand, it's very exciting, and I think we're really changing the way we're doing things.

I think the potential is great because as we were just saying, the complexity of some of these problems makes it very difficult to do detailed modeling that really fully captures all of this complexity and then integrate all sorts of diverse data type, as we were saying, and maybe predict the impact or predict the function of the variant, then it is very difficult to do that.

But by now, we have all of these data sets that exist that are looking at this from all sorts of different angles. If the goal is to really predict the impact of the variant, you know, these approaches are the way to go, right? Because you can really, it's not easy, but you can really think carefully about what data to use, what types of training sets you use, what specific question you're asking, and then develop some very powerful tool.

And again, this is already happening now, and I think we'll continue to improve. And I think that the exciting part of all of this is that this has the potential to really integrate and use many of the data sets that we've been generating, that we've been using, but in a sort of piece by piece way and not in a more holistic way. So I think that part I find is quite exciting as a new development.

What's less exciting is that people now are also using this without really thinking and understanding what it's really doing and what the implications are. And I think, I mean, maybe that's because I'm old school or I worry about how the technology is being used. But I still think that it's important to understand why it works and why it doesn't work and use also this information to start growing our understanding as human, of human biology and how all of this works.

And, you know, machines might be very good at predicting things. And in the context of health care and treatment and all of that. I mean, we might get to a point where, without really knowing -- and a lot of what we do anyway is based on intuition, right? And it's and so I think these types of tools will gain amazing intuition to solve very important and difficult problems.

But I think, you know, trying to also figure out how we'll use these types of tools to improve human knowledge and work to facilitate scientific discovery. Or without human, only machine knowledge, moving into a different type of knowledge and doing things without us. So that's the part I find scary and it's a bit more worrisome.

I mean, this is, I guess, in a high-level philosophical way. But this does translate more practically in terms of, you know, the kinds of use that I already see of these tools with people thinking that they can just use a tool and don't need to know or learn anything anymore. And then, you know, and I, I think that that's wrong because, you know, you will not know how to use these or develop these tools and machines if, yeah, if you don't understand the biology and then try to tune these approaches to that.

00:45:44 JP: That's a very important point, absolutely. And I think that's a very big pitfall that we are facing right now as a society in general. It's important to remember that we cannot just apply a black box to a problem and then just take the solution for granted.

00:46:03 GB: Yeah. But I think as a new generation starting into this, I think this is very exciting. If I'm putting myself in shoes of a new student, what is important to learn in this context is really quite difficult to predict, I think. Like, do you need to just focus again on learning this hardcore machine learning, or do you need to focus on understanding some of the technologies, I mean, I think it's changing, it's even more than I've seen in the past.

I think it's really, I think what we end up teaching and learning, you know, is going to change quite a bit. And it's- it's, yeah, both exciting and challenging time, but it's impossible to avoid at least keeping track of this to some extent. And I think making a conscious decision in terms of, you know, what to do in this space or not to do.

I mean, I think it's probably fine to be a biology expert with some understanding of some of this and then plan to work in collaboration with somebody who would have the reverse. You know, like when I was doing what I'm doing, it was a bit the same. You know, you had wet lab biologists and you had dry lab biologists and I was a dry lab and I worked with wet lab.

And I think, you know, a similar kind of collaboration I think is needed here where you I think you do need biology knowledge experts and then you do need machine learning experts. And I think the greatest progress will come at the interface of these two things. And I think it's hard to be a real deep expert in both. So I guess that would be my advice. Pick a camp and then make sure that you're talking with the other camp.

00:48:11 JP: I think that's great advice, yeah. Because you can't know everything about everything. That's just not possible.

00:48:21 GB: That's right.

00:48:24 JP: Great, yeah, very nice.

Conclusion

0048:27 JP: This was part one of our interview with Guillaume Bourque. In the second part, we will talk about his involvement in the McGill-Kyoto joint PhD, and how these kinds of collaborations are shaping the landscape of Genomic research. Stay tuned!