 Welcome, everybody, to our panel on biology and engineering life. I'm Nathan Benesh, founder of AirStreets. This is James and Hanu here, from Helix Nano and LabGenius, respectively. So in preparing for this panel, we were having a chat yesterday, and I was checking online what the status of biohacking was these days. And came across this article about this guy who was injecting himself with novel materials and being the exemplar of biohacking in a spare time. So I was curious in reading that and kind of considering why is it that now is the right time to be thinking about these topics, why is some of this possible, and what are some of the enabling technologies that underlie this thesis that we're talking about? Yeah, that's a really interesting question. So for as long as we've been studying biology, it's always been a question of understanding how we work, looking in nature and understanding that. And now there are these sort of emergent technologies that are enabling us to read DNA, to write DNA, to understand how those genetic programs work. And it's really the confluence of these technologies that are enabling these new developments. And it's not necessarily just on the DNA level, right? There's a whole bunch of other kinds of omics approaches that we're using to understand life. Right, and it's that fuller picture of understanding how life works at the molecular level that's informing these next steps. And so I thought we'd maybe ground the discussion in two areas that both of you have confidence in. And so the first one I think we should talk about is cancer. And how do you have some opinions as to why cancer is one of these important biological problems and is so intertwined with life itself and is a good way of kind of exploring why biohacking is kind of working today? So do you have some thoughts for us? That's right. Thank you, Nathan. And so cancer has been described by author Siddhartha Mukherjee in a book of the same name as The Emperor of All Maladies. And in many ways cancer is sort of a dark mirror of ourselves. It's our own biology turning against us. And in order to fight it, we actually have to understand the full picture of how biology works from evolution to DNA to all these control pathways of RNA and proteins and basically the whole deal. And I think throughout history, cancer has been the battleground where the sort of new and the most advanced technologies in medicine have been tested. And we sort of look back in the 19th century and sort of early 20th century where it was very much about sort of trying to just mechanically see how the organism works. The most advanced cancer treatments were radical surgeries. And it got to the point where doctors were essentially competing about who could remove the most tissue, like who could cut out and who sort of got into these horrendous fates for patients who sort of barely survived but sort of were missing most of their abdomen or limbs. And of course that actually ultimately helped because cancer was driven at the genetic level and then sort of was metastasizing and spreading throughout the whole body. And then sort of closer to our era of 1940s, 1950s was the era of chemotherapy, where there was the realization that you can target, you can kill cancer cells by targeting them with various toxic chemicals. And that became the most popular approach. But of course by doing that, you're also kind of indiscriminately killing other types of cells as well. And it wasn't really until sort of 70s, 80s as the picture of the role of DNA as being sort of the programming language of our bodies that sort of people started to be able to tell what actually makes cancer cells different for normal cells. And people identified some genes, some elements of DNA that were sort of mutated and were driving cancer and tumor growth in specific cancers like breast cancer. And that led sort of to the first, in the mid-80s, to the first targeted therapies, whereby hitting those specific pathways, specific things that the sort of cancer cells were expressing and reproducing, you could actually specifically kill cancer cells. And that led to therapies like Herceptin, or sort of a really blockbuster cancer drug from Genentech. And now we've sort of come even further. So now that our ability to read DNA, to sequence DNA, has gotten to the point where it's sort of of the order of $1,000 to sequence the entire human genome, sort of down from $100 million around the time of when the human genome project started. So sort of really a faster rate of progress than even Moore's Law. Now we actually have the ability to sequence a patient's entire tumor and sort of look at the full landscape of all the different mutations that occur and are sort of driving cancer growth. And we can target some of those specifically. But what I find really fascinating recently that first of all, people are trying to apply machine learning to those sort of data sets to predict which ones of those mutations our immune system is capable of targeting. And then actually taking some of those mutations, not just reading DNA, but also writing DNA, putting those mutations into a DNA construct and putting that DNA back into the patient's body to train their own immune system to go after those cancer cells. So this is a field of therapeutic cancer vaccines that's quite exciting. Or even more radical approaches that are now no longer speculation or research, but actually becoming clinically approved FDA approved therapies like Nomartis's Cumbria, where you're taking immune cells, white blood cells from a patient's body and genetically engineering them, sometimes quite extensively to go and target target cancer. So, and it's that sort of whole stack of technologies that we've had to put together to fight cancer. That's also now starting to find applications in other areas, which are actually quite surprising, which I think James is well placed to tell us about. Yeah, so some of the things you highlighted, which is one is sort of the intersection of biology and physical materials and how solving diseases not necessarily about just understanding biology, but actually getting the therapy or getting the treatment modality in the right place at the right time. And that also involves physical materials. And then the other idea is like this high throughput experimentation, which is how can you address a vast search space of possible therapies or possible underlying reasons why a disease actually occurs and use things like robotic process automation, which I know James, you're also using in your company. And so I think we should also talk about that and kind of consider how physics is interacting with biology and how advanced automation tools are actually allowing us to address some of these really difficult problems. Right, so it's a really interesting question and it's one around how do we fundamentally approach the engineering of biology? That's right. And that's a super interesting question because biotechnology up till now has really just been about how do you increase the production of existing things? How do you port processes from one organism to another? But how do you create things that have never existed? I mean, how do you even start? Like that's a super interesting question. And I think fundamentally this boils down to kind of the way in which we see the world, right? So all around us, like everyone here looks remarkably similar. We all occupy this very small island and genetic space but the potential of that huge space that could encode any organism is vast, right? So how do you explore it? How do you create new things? And human thinking is often tied up around cognition and how we interact with the world. So I mean, even the words that we use like manufacture, mannerism factor to make with your hands, we can't kind of adopt the same engineering principles when we start playing around with life. So the approach that we take at LabGenius really is around how do you build systems that can build themselves? And that's actually like a very natural approach. So taking evolution, the reason that everyone here is sitting in this room after 3.8 billion years, it's that process that can give rise to incredible complexity. And the beauty of that is if you can get systems that can teach themselves how to engineer life, you can be at the pub having a pint and you don't have to understand how they work. So the kind of approach that you take to do that and the way that that approach leverages these technologies that we're talking about is remarkably simple. Evolution consists of a few simple steps. You create genetic variability. You interrogate that variability to understand what works and what doesn't work. And then you propagate the stuff that works. Now, natural evolution is this incredible process, but there are fundamentally two problems with it. The first is it's super slow. It's taking time to get here, right? And the second issue is the solutions that it creates, they're far from optimal. It's like a local optimization routine. So, I mean, I'm not picking on you, but you have far too many bones in your feet, you know, your nasal passageways, they're not like functional. It's not optimal. Right, right. So the beauty of synthetic biology is it enables you to explore that potential sequence space without being tied down with the limitations. So we create trillions of unique variants of genetic designs, screen them in parallel, so you're empirically working out what work and what doesn't work. And the beauty of applying artificial intelligence to that problem is that you can actually learn that stuff so that when you create another trillion unique DNA sequences, you're picking out the stuff that works from the get-go, and that enables you to really accelerate that evolutionary process. And so suddenly you can start engineering biology right down at the nanoscale, and then you can look around yourself and you can say, what would the world look like if you could build with nanoscale components? And that fundamentally changes the way we interact with the world. Yeah, I think this search space idea is incredibly powerful, and for those of us who are machine learning enthusiasts, it's also exemplified by AlphaGo, where the best player in the world is actually just a local optimum in our actual knowledge of the ability to play go, and many of the themes that you're describing are actually very similar, where you no longer have to have rules about the world, but you're instead learning the rules automatically from data by just abstracting away a lot of this complexity. So even Mika Hakkinen is sort of a locally optimal F1 driver. He could be. We could probably evolve a far better one. That's our next project. But I just wanted to comment on sort of, I mean, James Nicely described sort of how the kind of engineering process that we're starting to apply in biology is actually quite different from the traditional approach to engineering, which is where you have some standardized components and you kind of rationally try to think about how to put them together in the best way. And it's very different from sort of taking millions of designs and then sort of throwing them into your sort of experiment and seeing what comes out and then using an algorithm which we actually, the internals of which we probably don't really even understand that sort of use these features that we can't sort of necessarily interpret very well to then analyze and then design the next experiment. So we're kind of decoupling quite a lot of our understanding of the process from the process itself. And I guess we have to get comfortable with that. And I guess I want to ask James, are you comfortable with that or how did you become comfortable with that? I think it's just like, that's probably at the heart of like how I like to think about the space in them and what drives me in the sense that right now we have to as humans understand the products that we build and that creates a fundamental barrier to the complexity of the stuff that we can create. So if we want to create stuff that is more complex than we could actually understand, which is I think an absolute necessity for our species, we have to embrace using these self-learning systems. Yeah, I mean, it's this interesting paradox that we impose like a higher rigor of explainability on things that we build versus ourselves in many ways. If you ask somebody about how the reason about that the fire in front of us is actually a fire, like at the biological level of how neurons fire, they're completely capable of doing that, but we're imposing that level of rigor oftentimes on machine systems. But I think towards your point, like you have to kind of liberate yourself outside of that bound to try and discover that new solution. So do you have some examples of what you might have discovered with your process by uniting this combination of biology plus physics to modify materials? Yeah, so we've actually deployed this technology across a whole range of different areas. So we've helped traditional biotechnology companies do what they're doing today, so engineer new biological molecules for therapeutic applications. And then we've actually started exploring some really interesting spaces, even outside traditional biology. So we'll work with material science companies and we'll say, what would you do if you could engineer matter at the nanoscale? And they get super excited by those kind of applications. So we've had a number of projects financed by the UK Ministry of Defense around making lighter body armor, novel attesives, novel nanocomposites. And these are areas that you wouldn't naturally associate with biology, but because biology gives you that control of the nanoscale, it's a really powerful tool in these spaces. I guess sort of to comment on some of the Helix nanowork, I mean, so to date our approach has been not quite as radically sort of high throughput and powerful as what James is describing, but what we've done is, one of the core problems we've been trying to solve is actually when we write DNA, how do we get it into the right place in the cell to actually do the job? So our DNA lives in the nucleus, but the nucleus is very well protected from all the viruses and other things that might try to get in there. So if you actually want to get some DNA in there to either edit the genome or to cure genetic disease, you have to overcome that barrier. So what we've done is we've sort of taken some inspiration from the millions of years of viral evolution where viruses have evolved sort of tricks to hack the nucleus, but then we've sort of thrown the virus itself away and engineered synthetic proteins and sort of synthetic cellular machinery that sort of mimics some of those things that the virus is doing, but in a way if the virus is a bird, we've been sort of trying to build an airplane, but as we are optimizing that, that system we're also moving away from that sort of designing by hand and having an algorithmic design that to generate sort of many possibilities, many variants of this system and then testing those to see what we get. So in a way we're sort of bootstrapping a system that then will enable us to better manipulate biology and sort of to improve itself better as well, but at the biological level. But by removing a lot of the downsides that have historically prevented those kinds of approaches that we're actually making into the clinic, right? That's right, that's right. So yeah, I mean, a lot of the traditional ways to treat genetic diseases involve infecting the patient with sort of the disabled virus essentially, but you can still have a very strong immune response against the virus. So as well as the virus has its natural limitations. So it can only bring in sort of a certain amount of DNA and we don't have that limitation. And so when you think about our ability to control and evolve new biological systems, do you think there's like some things that we should be allowed to do and some things that we shouldn't be allowed to do? We talked about the malaria mosquito as an example where it's obviously a bad thing that they exist and I think we would all agree that they shouldn't be here, but there's other situations where that might not be the case. And so how do you put in place paradigms that allow people to reason about what's okay and what's not? I mean, I guess the default assumption in biology is that whatever can happen will happen. I mean, there's sort of enough complexity, enough variability for sort of the organism to try out all possibilities. So I think what you're referring to, Nathan, is that there have been proposals to use a technology called a gene drive to actually eradicate malaria mosquitoes and people are already experimenting with this. So what this is about is so when in normal sexual reproduction, a new gene is only passed to basically on average, sort of one in four descendants if only one of the parents has it. But gene drive is sort of a new way to ensure that all of the descendants of the organism get that gene. And people have designed a kind of a suicide gene that makes malaria mosquito males infertile. And there are now experiments to sort of deploy this locally and it's been shown to spread through the population like wildfire, but the concern is what if that gene actually doesn't even remain restricted to that species but jumps to another species or mutates, which again, biology is very good at doing. Which has happened. Escaping the sort of evolutionary dead end. So the question is, are gene drives that potentially give us the ability to manipulate entire ecosystems, not just single organisms? How safe are they and what can we sort of put in place? I mean, I think it's quite telling that the inventor of the gene drive, Kevin S. Belt, actually in a recent New York Times article said that he actually has serious concerns about their use. Although he then also said that if he lived in Sub-Saharan Africa, then he'd probably be inclined to go for it. So it is a very tricky equation. I mean, on one hand, you have this sort of enormous potential to eradicate the carrier of one of the deadliest diseases in the world and on one hand, sort of the unforeseen consequences that that technology could have. So I don't think we're actually very good at balancing those kinds of trade-offs as a species. Do you disagree, James? Yeah, so what I think is interesting is really the power of biology is this double-edged sword, right, where you can take one copy of a single organism and maybe terraform a planet. Like, that's so powerful equally the same organism could maybe wipe out another species. So this is where biology differs from traditional tech. It's the emergent properties of these complex systems. It will make some A so powerful but B so difficult to predict an engineer. And it's that natural tension, which is like both the threat but also the opportunity. Well, like, how do you guys evaluate the adoption cycle for many of these technologies? You know, a recent one is a nature paper that shows that you can encode information into DNA. So instead of storing it in AWS and cold storage, you would instead write it in DNA and you'd have your image files, whatever, in a lab. And if you wanted to retrieve them, you could just sequence it, output binary bits, and then re-synthesize your image. And that sounds fantastic. And in a world where data is exploding everywhere, we need to discover new modalities to store it. But, you know, it strikes me as that the adoption curve is probably going to be a bit slow. So how do you guys actually reason that when you're considering, you know, what you should be working on as a business? Yeah, so I think that's a super interesting question. And the way that we kind of think around that is what's the long-term potential of the technology? And obviously that's infinite. But then how do you realize value in, you know, three, five, 10 years? I think the way we kind of look at that is there are markets and opportunities that we can target today and make a real impact in whilst we're building out a platform that will ultimately disrupt this much larger space. You know, I think that's really, really sort of a good way of putting it. And as it's actually in Silicon Valley, the sort of common recommendation for any pitch that you have the moonshot, but also the sort of practical near-term application. And our moonshot, obviously, is this sort of freeing us from the constraints of our natural genome. I mean, giving us the ability to insert into our bodies any gene we want that can prolong our lives, prevent disease, give us new abilities, sort of all kinds of beneficial things. But then the short-term opportunity is to take some rare genetic diseases like hemophilia, where those patients have a terrible quality of life, have to sort of take synthetic biologic drugs every other day. And if we can give them back the ability to produce the sort of missing blood clotting factors, for example, hemophilia, back to make those molecules in their own bodies, I mean, that's completely life-changing. So I think that's both from a human perspective and sort of a financial perspective, a great opportunity. Yeah, yeah. So on that note, gene synthesis, data storage, machine learning, engineering. It's going to label us to solve amazing problems and discover new solutions that we've never even imagined to exist. Super inspiring. James Hanyu, thank you so much. Real pleasure. Thanks, Nathan.