a16z Podcast - a16z Podcast: On the Genomics of Disease, From Science to Business

Episode Date: October 18, 2016

Once we sequenced the human genome, we'd know the cause of -- and therefore be able to help cure -- all diseases... Or so we thought. Turns out, 20,000 genes (and counting) didn't really explain why disease occurred. Sure, some could be explained by mutations in a single genome, but most, like cancer, are too damn complex. And while the focused, singular approach to understanding disease did yield some useful therapeutics, it's now reached its limits. It hasn't helped much on the diagnostics (and early detection) front, either. That's where a systems approach to bio comes in, drawing on machine learning techniques as well as a sort of "Moore's Law" for genomics that's driving costs down, and fast. We're now focusing on the 99% of the genome that hasn't really been understood yet in terms of how they affect the human body and disease. But what will it take for such an approach to succeed? For one thing, it involves building an applications layer on top of the sequencing layer -- so can we borrow lessons from how the computing industry (from chips to apps) evolved here? What are some of the constraints unique to the healthcare system? In this episode of the a16z Podcast, Freenome CEO and co-founder Gabriel Otte and a16z bio fund partners Vijay Pande and Malinka Walaliyadde (in conversation with Sonal Chokshi) talk all things genomics and disease from science to business, also covering recent news like Illumina to what's next beyond human genomics to future trends. Including what the ultimate, Elysium-like magical diagnostic machine is (hint: the magical is mundane!).

The content here is for informational purposes only, should not be taken as legal business, tax, or investment advice, or be used to evaluate any investment or security and is not directed at any investors or potential investors in any A16Z fund. For more details, please see A16Z.com slash disclosures. Hi, everyone. Welcome to the A6 and Z podcast. I'm Sonal, and I'm here today with Gabriel Ott, who is the co-founder and CEO of Freenome, which is a company that aims to cure cancer. through early detection using machine learning and other software techniques on DNA. We also have our general partner, Vijay, Ponday, who heads up our bio fund. And we also have Malinka Wali. Oh, my God, I'm going to put you your last name. Walla Ljadde. Perfect.

And that's so embarrassing because I'm Indian too and I really shouldn't know how to say your last name. It took me a while, sir. It took you a while to say your last name. That's good. It was like five probably. Well, great, you guys. Welcome. We're talking about machine learning techniques brought to bio and what that opens up for us

and what some of the challenges are as well. So just to kick things off. Yeah, I mean, one place to start is that, you know, ever since the Human Genome Project, people have been talking about all these great cures and things that will come from it. And what is different now? You know, why are we expecting to see something now? I think one of the biggest realizations of the Human Genome Project is that the human

body and the genome is much more complicated than we thought. Everyone thought we would know the cause of all diseases and therefore we would be able to cure all diseases. As soon as we realized that 20,000 genes just were not enough to explain while all these diseases occurred, it gave life to all sorts of new fields of biology that have been built around understanding the genome. What's different now as opposed to what was happening in 2000 is finally the technology, the machine learning techniques as well as the hardware supporting that has matured to a point where we don't have to try to manually figure this complicated system out by ourselves.

we can do it in a computationally aided manner, which is greatly accelerating research in this field. Tell me why that makes such a big difference because, you know, my image is of genomic scientists sitting in a lab with like pipettes and just sort of doing like their studies or whatever it is that they do that's not in a horror movie. Sure. I think in the 80s, 90s, you could get an entire PhD studying a single gene. The simplest way of understanding genetic diseases is a point mutation in a single gene that causes a certain disease. And Huntington's is a great example of that where a single mutation can cause a tremendous amount of neurodegeneration at a certain age. But most diseases we found out weren't like that. And in fact, most diseases were so complicated that we didn't necessarily know the root cause of it genetically speaking. So like just for a very simplified analogy, if you think of the letters of the alphabet and just like words are created by a combination of letters, you now have to actually hone in on a more overlapping set of variables. Is that overly simplified?

That is actually pretty good encapsulation of what's going on. There's an entire field called Systems Biology that's been created around understanding how genes interact with each other. And yes, there are the A, Cs, Gs, and T's that's the basic code of life that creates a lot of these genes. And any changes in that code could affect cellular function and therefore physiology level function. But we realize that it's often a concerted change that results. in a serious disease as opposed to a single change that can cause propagation down to the physiology level. You know, there's an irony here, which is that in the old days, people used to do things

physiologically. You'd look at tissue, and that it was a major advance of molecular biology would go from tissue to an individual protein or an individual gene and would go after that. And only now that we're having this nuance view that actually the physiological approach actually makes sense and that the system's approach comes in. The other key part of what's now is that human genome project costs $3 billion, and that's a pretty hefty co-pay for someone to have to shell out. It's only now that genomics is starting to get actually to the point where you could imagine putting it in production where it would cost thousands of dollars or maybe even hundreds of dollars. Why is that, though, I mean, there isn't a Moore's Law like there is in semiconductors.

So what's the equivalent in genomics? Yeah, there is. I mean, essentially it follows a Moore's Law-like curve. The intriguing thing is that it's not the same hardware that is doing it behind it, but it might be the same. same sort of will to power. And it's beat Morse law by several magnitudes. Yeah. What is the it, though? Like, what is the thing that's accelerating the way Moore's law and driving down the cost for biology? The dominant player in the sequencing machine space has almost entirely by itself been driving down the cost of sequencing over the last decade or so.

And an easy way to think about genomics is there's a sequencing layer, which is the companies that make the sequencing machines like Illumina, which is like sort of Intel and chips. And then there's the application layer, which are companies that are using this genetic data to make clinical diagnostics. And that's sort of like Microsoft, whatever, software companies building on the hardware. So it's sort of like the equivalent in the semiconductor industry is the hardware and chip makers. And the sequencing layer is the people who are doing the chips, the fundamental substrates. And the application layers are people who are building the software applications on top of it that we can use. Okay, that's super useful.

So you guys have definitely answered why the shift is happening, like a more systems approach, things are more complex. That's kind of obvious to me. The complexity pretty much invites computer science as a point of doing things that humans cannot calculate. But what are some of the manual versus automated things that you guys have described as then and now, like what's changed in the lab and people's practice? So a lot of my PhD work was in a field called computational biology, which is in some sense a halfway house between, completely generalizable machine learning approach to answering these questions and just looking at single genes. Computational biology is really a field that's designed to use computational tools aided by the biological knowledge of the scientists that's wielding those tools. So some of the things that you can do in computational biology, for example, is take the DNA data

that's coming from an Illumina sequencer, for example, tremendous amounts of data. you're going to have to go through essentially prepare that data before you can do anything with it. That step is called DNA alignment. And that used to take days. Now there are tools out there that can do the alignment process in five minutes. So that's some of the things that that field has contributed to in the last 15 years. Where I think computational biology could use help from things like machine learning is making sense of what that amount of data is actually same. about how we understand disease and how we understand human health.

And that's something that manually is very hard to do because, let's say, you can ask what is the RNA expression level across those 20,000 genes. That doesn't answer the question of which genes are relevant for determining whether somebody has a certain disease or not. It only tells you the raw values across those 20,000 genes. Then the human has to go in there and figure out what are the relevant parts. Now with machine learning, we can start using computational tools to answer that question of what are the relevant parts of this data. Yeah, in particular, the old paradigm of looking at snips makes sense because that's what a human being can do.

You know, we think even like back to the... By the way by snips, you mean the DNA. Yeah, single point mutations. So almost like back to thinking about Mendel and his peas, you know, it's something where a human mind can understand that. But that's where it stops. And actually a lot of genomics today is just rows and rows of peas. looking at papers, looking for these mutations, and then using that to assign. And, you know, biology is most likely much more complicated than that ability that's limited

by human minds. And so where machine learning comes in is that it opens up the door to analysis that really human beings couldn't do, much like computation has done more broadly. Exactly. It actually reminds me of the podcast you guys did with Herman Nirola, and we were talking about simulations and improbable. Because what's interesting to me is this is not just the disintermediation of existing tasks that people do are making it faster. It's about opening up things that nobody could

ever have done before because of the way a computer works. Building off ML techniques allowing us to take and use much larger data sets than was previously possible. In diagnostics generally, people have been looking at very specific parts of the genome in the past and trying to make interpretation space of that purely. What Gabe's able to do is take an agnostic approach and look across the entire genome and use lots and lots of different points of interest, which obviously It creates a lot more data to work with, but because they use these very novel ML techniques, they're actually able to accommodate those massive large data sets and make more accurate and more useful interpretations.

What's happened is there are certain genes, P53, K-Raz, H-Rs, EGFR, that are sort of the the usual suspects. That's a very nice way of putting it. I'd use something way worse, like assholes, the assholes of the genomics world. But anyway. And so the research has largely fixated. around understanding how these mutations affect the cell, affect the body. Where less research has gone into is what other aspects of the genome are actually involved

in that process. The genes that I mentioned, we're really talking about less than 1% of the entire genome. And so there's this 99.99% of the genome that hasn't really been understood in terms of how they affect people's bodies and how these diseases come out from other parts of these genomes. How do they manifest themselves? So this ties back so interestingly to the points you guys you're bringing up earlier about the systemic approach, why the singular focus never worked as much, which worked before, doesn't work as much now because of the complexity of this. Or just at least it's had, it's reached its limits. It's reached its limits. I mean, to be very clear, there's been

really, really great therapeutics that have come out from understanding the cancer in this very focused way. What sometimes happens when you are very focused, though, is you missed a bigger picture. And what we're realizing is with respect to things like detecting cancer early, that focused approach is not working. There are aspects of it that's just much more complicated than we thought. The other thing is that there's a differentiation between diagnostics versus therapeutics. So therapeutic being focused, that makes a ton of sense. Previous diagnostics were either something you could do by imaging or something that you do by biopsy. And biopsy, you can do genetics there, but you're not going to do biopsies prophylactically. I don't think you want

anyone like taking tissue from all of your major organs, like once a year, that's a disaster. Really, the other key advance here in the cancer space and genomic space is the fact that blood has so much genomic information, even from things like tumors. And that opens the door for these new technologies to come in. I'm going to ask a really obvious question, though, which is why is cancer so hard? I get that it's a complex disease. Yeah, I think there's a couple answers. One, it's not a disease. It's really many diseases. Number two, actually the finding is the one I was studying biology many years ago, you know, my knee-jerk reaction immediately after reading this stuff is like, I'm amazed it works at all.

So the fact that apoptosis breaks down or there's some somatic mutations that mutations happen, that's actually not that shocking. I think the fact that actually our body does such a good job of maintaining things to me often sounds even more surprising, but, you know, we can do more to help it. In addition to machine learning, be able to come up with the quillin of new types of biomarkers that people couldn't come up with, there's a learning aspect here of machine learning, intriguing that as you get more data, you get better. And that's something that's really not like any other test where, you know, a lipid blood test doesn't get better as you have more patients there.

And that's, I think, a really intriguing aspect, especially as we wanted to detect cancer earlier and earlier, where there's fainter and fainter signs, and especially for a wide range of cancers. The question is, why is it important? Because for some people, it might seem obvious, but it really isn't obvious. The best drugs that we have today, to treat cancer called immunotherapies only give you about 30 to 40% chance of five-year survival in treatment. These are the best drugs that we have. Chemotherapy and radiation give you less than 20% chance of survival. On the off chance, we get lucky and we detect this disease early for whatever reason.

That chance of survival goes up to 80 to 97%. Oh my God. That's a huge difference. We're trying to systematize that early detection so that it's not a chance event. anymore that we can do this for everyone. It's not about just diagnostics curing the disease. It's about the early detection and early therapeutics that enables the maximum chance of survival. With that in mind, I think a lot of us like to think that it's going to be the next greatest drug that's going to cure cancer. It's going to be that silver bullet. There isn't really going to be a

silver bullet in my mind to treating 100 different diseases simultaneously. I'm glad you brought up because I take it for granted that there's this question of why it's important, so why is the detection part early so hard? Mostly because we're going against the paradigm of medicine the way it's been practiced for the last 2,000 years. The way medicine has been practiced is we practice we practice what's called a symptomatic medicine, symptomatic detection. Essentially people have symptoms. They come into the hospital. They get diagnosed based on their symptoms, and then they get treated. The unfortunate thing with diseases like cancer and Alzheimer's disease and various other age associated diseases is that by the time you show symptoms, it's often too late. So we have this

paradigm of just studying symptoms and trying to match diagnoses to symptoms. But in order to beat these more complicated diseases, these more virulent diseases, you have to find a way, you have to build a technology that can detect these diseases before the human being shows any symptoms. And that often correlates with just very, very difficult signals to find within the body. Yes, there are signatures early on that indicate whether there is a tumor or not. But what hasn't been solved yet is exactly what are those signatures that allow us to detect those diseases and what are the best signatures for us to detect them most accurately. When you say early detection, we should separate risk from diagnosis.

And so what about these Bracca tests, you know, breast cancer test? There was article about Angela and Jolie doing this some time ago. So that's different because risk tests just tell you your chance of getting cancer. They don't actually tell you if you have cancer right now. And so if you're at higher risk, that probably means you want to get tested more frequently, but we don't have good methods of testing and diagnosing you with cancer early right now. Yeah, that's a great point. We're not doing any kind of prediction.

We're doing detection. it's not really empowering to the consumer to know that they're going to have 30% chance of getting cancer sometime in their life time. Other than the fact that maybe if they're at hard risk, then they should get a test. I would say speaking on behalf of Angelina Jolie, my best friend, that, you know, I think for a lot of women, when you do have a very strong family history of breast cancer, there's actually something very importantly empowering about being able to proactively do that, which is, I think, the point of her op-ed in the New York Times and making that choice, whether it's the right choice or not, it's an individual one. But I hear your point that you're really talking about the fact that at the end of the day, when it comes to the mass patients in the system, it's really being able to detect versus predict that's important. Well, I imagine this is another choice for women to be able to do instead of doing something so significant as a radical mastectomy. Right. Exactly.

To be able to just know whether there's an issue or not. Right. And early. And it's obviously not a problem that just touches women. I mean, cancer touches everybody. Absolutely. I mean, I think one of the most inspired. Firing things as I've been building a freedom is hearing the personal stories of not only the employees, but my friends who have all been touched by cancer in some way, shape, or form.

My grandfather, who helped raise me as well as my dad, both have cancer right now. It's a story that we hear all the time from everyone is, you know, cancer is traumatic for everyone involved, not just a person that has the disease. And if we can somehow make that better, somehow give them a higher likelihood of survival, We're not just affecting those patients. We're affecting their loved ones as well. Yeah. This is a place where technology can do a lot. And I want to hear your guys' thoughts on what are the trends in this space that are being brought to bear on this severe problem? You know, the major news in the genomic industry this week, which is the largest genomics company of any kind at the sequencing layer, about two days ago had a 25% drop in their market cap, which definitely send waves in the industry. I was really puzzled because Illumina effectively is a monopoly in the space of DNA sequencing. And in a lot of ways, rightly so, their technology is very good.

We have one in-house that we use for our analysis as well. I think there's a chicken and egg here, which is that they're producing, in a sense, the hardware, and the software needs to catch up. T.J. Watson famously talked about how, what, there'd only be 100 computers in the world. And at a certain age, that made sense. but then the software caught up and now everybody has computers. I don't know when it's going to get at the point where we each have our genome sequencer at a house or something like that that might take a little while, but not actually crazy to imagine by any means.

Just interrupt you for a second. Why would we want that? Is it just so we can personally sequence your... Yeah, you can imagine like it's a little off color, but you know, you wake up the morning, you use the toilet. It sequences everything right there. Oh, my God. That is disgusting. Yeah, yeah.

But like, that's a lot easier than like taking blood. There's lots of possibilities that you could imagine. in the far future, you know, just like it seemed outlandish to have a computer in your house. In your pocket. Far less, or in your pocket. I mean, that just seems ridiculous. Like, who would need that? And you're right.

Okay. That's a fair point of view. But, you know, you wouldn't need the computer in your pocket if you didn't have something to do with it. If you didn't have the software. That's a key part here. And I think what we'll expect to see more broadly is that prevention in healthcare used to be like eat better and exercise and sleep more. You know, and there's real limits to what people can do there. but when prevention is catching diseases early through things like genomics,

whether it would be cancer or other areas, this is something where I think we can see medicine really radically transformed. I don't know if I want to expand on the tall analogy. Well, it wasn't an analogy. I was just saying it straight out. If you've seen the movie Elysium, they have this magical diagnostic machine that looks really cool and people like this is a diagnostic machine of the future. I actually think the magical diagnostic machine of the future is your bathroom.

Really? Yes. Very lucky. Okay. I'm willing to buy this. I just don't want to talk about it anymore. Sure. I can move on. And I also totally agree. There haven't been enough useful clinical applications expanded in the market. Illumina keeps selling these sequencing machines, but if people unbuilding applications on top of them, that isn't good for either party. If you think to the 90s, Wintel was a thing, you know, Windows and Intel together core marketing. And that took both of those companies to new heights. Something like that needs to happen in the genomics world as well. There needs to be a window, a Microsoft, in the genomics world, and a lot of people think that will be in cancer. So clearly, the sequencing layer is just one level.

We really need to focus on the application layer. And not just that, they're layers on top of each other. They're very symbiotic in relationship, almost, to use a biological analogy. And speaking of that, the broader ecosystem of all these players, like, you know, big companies, startups, what happens when you have a big player like that? Because frankly, you only care about a monopoly being anti-competitive if it's preventing consumers from benefiting, not other competitors. And as long as consumers benefit, why do we care? Like, what's the, is there, is that a good or bad thing in the ecosystem?

I mean, one good thing would be is that because of this data network effect, the ability to make a better product for consumers is obviously very favorable. The question is in which ways are they good and which ways there's a poor. And, you know, there's different aspects of monopolies here. One is a data monopoly. Another one is the sequencing monopoly. On the sequencing side, there are several other companies doing sequencing and sequencing using different technologies and different means. So you never know what's going to pop up. I mean, the dominant computer companies from the 70s aren't necessarily the dominant computer companies now.

Exactly. And so there are changes and evolution can be very positive. I mean, I want to hear what the opportunity is for startups are, obviously, like why you would even try to build a company in this environment. I think what, like, Luminah have done well is not necessarily preventing innovation built up. on top of their technologies from happening. I don't think that's always the case. There have been cases where companies are trying to be protective about their technology and how it's used, but by and large, we have been able to do our work without being disturbed.

And I think that's very good. I think where there are some worries in my mind is my PhD work in computational biology, I've never touched a machine that's not an alumina machine. There's this single technology that a lot of companies are basing. their technologies on top of, and if history tells us anything, dominant players don't remain dominant players indefinitely. If there are so many companies doing new, innovative things on top of a single platform, that may not be around in 10 years. What does that look like? Yeah, although I think that's common. I mean, in the computer analogy, there's lots of software

that was built on top of operating systems that we don't use right now, but then they get ported to new systems. Right. People find a way of making things interoperable. They find ways of adapting it. It's not so clear cut, but it's not like the work, I mean, frankly, I would even argue that some of what you're describing is really a commodity layer. Yeah. And it's almost kind of pointless who's labels on it at the end of the day. And so you kind of only care about the real value, which is the software applications, the data, services and those benefits. And interestingly, Lumina has started realizing that as well and is itself building applications. They bought a company called Varyana, which is a NIPT company. What's NIPT? Non-invasive prenatal testing. Oh, great. I like that idea. Yeah. You know,

They have a cancer company called Grail. There's a company called Helix. So they have started participating in the applications layer, which is interesting. And I'm curious to see how that works because it's sort of, I mean, if you draw this analogy that it's like Intel building software apps, perhaps they could do, but it's unclear. Actually, it's like any big company making a shift to a whole new type of business. It's like an on-prem software company becoming a software as a service company. It's like a hardware company becoming a software company, a non-tech company becoming a tech

company, which is happening everywhere. Yeah. It's a fascinating wave that it belongs to. putting aside all the issues of the monopoly and everything else, what are some of the commercial challenges in the space? Like, I mean, why hasn't it taken off more? You're totally right. There's been a lot of excitement and interest in genomics, both in the application and sequencing layer. One would you even argue hype. Yes, hype even potentially. And we haven't seen the output of that really. The largest public company on the applications layer is this

company called Exact Sciences, which is about roughly $2 billion in market cap, which is small relative to how much interest has gone into it. And I think it's, mainly because reimbursement has been very difficult. That's why you think. Yes. So the business model aspect of it, not even necessarily like any kind of technological limitation. No, it is a very business model-specific problem because the way these things work is

you need a provider. And actually, a lot of it comes down to the three-party system we have in the U.S. You know, you have the patient. Democrat Republican, no, just kidding. You have the patient. You have the provider or the doctor, and you have the payer, which is insurance company. and the patient doesn't really have any say in what gets prescribed to him or her.

In order to get a test sold, you need to convince the insurance company to pay for it, and then you also need to convince the provider or the doctor to prescribe it, and that is the only way this happens. The hardest part in this industry has been convincing the payers or the insurance companies to cover the cost of these tests. And maybe this is a little harsh, but I think they're being a little short-sighted insurance companies. they don't invest in things that have an ROI window of more than two to three years. And if a lot of these genomic diagnostic tests, you really realize the value of your investment

in covering one of these tests in more than three years. Yeah, exactly. This is actually so interesting because it's the same problem that happens, a lot of high deductible plans because you distribute risk and you're locked into a single insurance company, then you all see the payout. You're all inlined and incented because it's coming to you long term. But when you have a system where everyone can move around and switch and change and whatnot, you don't get this incentive for this long-term payoff.

Exactly. And that's been a major issue. In fact, I think in single-payer systems like the UK or outside, we might see much more adoption of some of these genomic tests. But yeah, that's been one of the main constraints. A is getting insurance companies on board willing to reimburse. There's a company called a Shurik's call in mental health. They were making $60 million in revenue last year, but they were only getting 20% of their test reimbursed.

That's crazy. They were only getting 20% of their test paid for, disposed. the fact that these tests were providing value. And so that, you know, that challenge has been very hard to overcome. So how do people overcome it? Like, is that going to change anytime soon? The main pushback that a lot of insurance companies have been giving these genomic diagnostic companies is that they are too expensive and they are not accurate enough. And in all honesty, I think that's somewhat true. Given the technology you have today, we're very good on accuracy, but they're usually in the thousands of dollars and they are useful. But at that price point,

is it useful enough, is a question? Right, to make that ROA. Exactly. So what needs to happen is for these tests to get more accurate and cheaper. And in order for that to happen, we need some of these new startups with these new technologies. You know, we've talked about cancer as an application, but what other applications do you think there are for genomics, clinical or otherwise? I think there's a lot of interesting opportunities because clearly people aren't the only organisms on the planet. There are a lot of interesting things to do with plants. And obviously, you can test plants in a way that you would never test with people or animals, and you can make better crops or just find bitter conditions for them. And even livestock, too, is a very

natural approach that you can understand better which feeds would help them or other aspects of improving your herd and making farmers jobs easier. And this is not like when people talk about eugenics like cloning and designing. It really is eugenics for cattle. The thing is people have been doing eugenics for cattle ever for the last 5,000 years. Actually, you're right. And for plants too, for that matter. Why do we sort of resist that idea? I mean, dog breeding. Look at all these people with their cute little dogs and their purses. Like, I'm pretty sure those dogs have been bred. Yeah, they look a little different from wolves. So what are some of the other interesting trends or opportunities that you guys think are ahead just in this space in general? Personally, one of my favorite

topic is proteomics and like all the mass spec work that's happening. Yeah. There are a lot of interesting molecules in blood. You know, there's proteins, there's RNA, there's lipids. Proeomics is particularly intriguing because mass spec is so sensitive and also doesn't require and therefore it doesn't require much samples. And by mass spec, we mean mass spectrometer is the instrument that actually measures. Yeah, exactly. And the amazing thing about is that in principle, you should be able to just prick your finger, put it on a piece of paper, let it dry, mail it in. And mass spec should be able to get something from that. Whether that's useful or not remains to be seen. But it's a great place also for machine learning to come in because the signal that comes out in mass spec does need a lot of

interpretation typically, but yet it's very data-rich. So another opportunity for it to continue learning and actually benefit the NL-T-B. Yeah, absolutely. An opportunity for it to learn or to fuse with other data sets. Vichie talked about novel applications of genetics outside of humans. I think there are also novel applications of genetics within humans as well. We talked about different types of applications. Non-invasive prenatal testing and infertility is one major area. In fact, that was the first to make it big in genetics and it's gone a little saturated

cancer is another one, consumer is another one, but there's entirely new verticals of applications. Two very new ones are mental health. Apparently, it's possible to, using your genetics, predict what type of psychotherapies and drugs work best for mental health conditions. There's a company called Oshurix that was just bought by a myriad for half a billion dollars. There's an entirely new application that's, again, very nascent in infectious disease. And there's lots and lots of other new things that we haven't discovered. I'm very, very interested in seeing some of those come about. Yeah. And one of my other obvious favorite topics is CRISPR and gene editing. But we'll save that for another podcast.

We should keep in mind that just from our knowledge perspective, we've just scratched the surface on how complicated this entire system is. There is the DNA, there is RNA, there is protein. But then there's all sorts of variations within those systems. And then we haven't even begun discussing spatial orientation of, for example, the DNA within the nucleus and how that can affect. So you mean like where it's placed? Yeah. So, you know, there are DNA when it's sort of free floating within the nucleus is organized in a very structured way, the 3D confirmation. So oftentimes genes that are transcribed together are actually spatially close together, but they may be on separate chromosomes. The cell knows how to bring those together so that there's maximum efficiency around that

system. And we haven't, we don't really even have good technology to really understand that yet. I think we really have to pause for. moment and remind ourselves of that that we're not at the end of the computing revolution by any means. I mean, clearly things are jumping by leaps and bounds. You're talking about deep learning and AI. But in bio and computer software, we're at the very, very beginning. You're right. I think we forget that. And not to get all crazy sci-fi, but what you just described reminded me that a cell is essentially a mini-computer in and of itself. And if you think about the human body as this whole system and computer or a set of computers, you literally have like this inception or it's like. like computers inside computers, inside computers. It's kind of fascinating. It's really good networking, too. Oh, yeah, a lot of networking, too.

Yeah, absolutely. So I'm really excited to see what new technologies come about that's going to enable us to study things that we couldn't before and what kind of new innovations that will create into health space that we're not able to even think about today. Well, that was really interesting. Thank you for joining the A6 and C podcast. Thank you. I'm just really glad you didn't bring up any more analogies about bathroom and poop.

That wasn't an invitation to add more. Also, just like the mirror, it can scan you in principle. You can have stuff. Your toothbrush could be, you know, Bluetooth-enabled.