With advances in biological and genomic sciences accelerating at an ever-increasing pace, what does the future hold? If biological progress is indeed advancing more rapidly than Moore’s Law, as many assert, what are the economic and societal implications? In this video, Alex Lash, biotech editor at Xconomy, interviews Drew Endy, bioengineer at Stanford, about biological processes.
Lash: Well thanks for having us here, thanks David. As you mentioned, I’m the National Biotech Editor of Xconomy, no relation to Techonomy except in spirit.
You heard about Drew. You also heard, just previously, a bit of an intro to synthetic biology, which Drew certainly knows, perhaps more than anyone on the planet, about. There was one thing that you just heard about at the end of the talk that I think is a good way to start our talk. And that is, with all of these tools, perhaps even Autodesk supplying them, there will be, as many people believe, a democratization, as there has been on the tech side, now on the biology side. That’s certainly something you’re working on.
Why don’t we jump into that and ask a bit of a provocative question: How much democratization should there be in the tools that will help create the biology of the next fifty years?
Endy: And you want me to answer that? I don’t know what to make of that verb. I think dissemination and promulgation are interesting. When I hear “democratization” I think everybody will vote on whether not Andrew should be making a personal virus for somebody.
But I do think the question brings up a couple points. When I think about the future of biotechnology, the two fundamental trends that seem ripe for creating laterals are, number one, distributed manufacturing. Biology is a way of making things which can make things almost everywhere on our planet. And right now we’re in a point in time in our civilization where we have a lot of centralization of manufacturing. So think about what about your bread machine can make. Well that’s a reactor with a bunch of yeast, Saccharomyces cerevisiae, which is a workhorse of biomanufacturing. Samsung should be thinking about the content in their bread machines, not just the hardware.
The other fundamental trend, which, to play off of democratization, is coordination of labor. There’s something about a democracy that is bringing people together to have at it. And when I think about synthetic biology, what it really reflects is gaining access to a distributed manufacturing platform, but secondly, coordination of labor. And this is a relatively esoteric topic, in that it depends on things like metrology, how you decide that a meter is a meter, and reference standards, and giving and getting. But it’s the technology platforms that allow people to coordinate, and if people can coordinate their labor, then all of a sudden you make things that are impossible. My favorite example from 2,000 years ago is the aqueduct in Segovia, Spain, which is made from standardized rocks. And the significance of the rocks is that they’ve been coordinated—they’re manufactured in a way that somebody could make them in a quarry and another person could assemble them. And 2,000 years later, if one of them crumbled, we could rebuild and sustain this. Now by myself I could never make that aqueduct, but if I could work with other people…. So democratization, for me, unpacks and re-directs into distributed manufacturing, meaning everybody can manufacture, because everybody’s got biology. And then it also is about coordination of labor. The last forty years of biotechnology have had a lot of sharing, but not often sharing of technology.
Lash: So coordination—we can perhaps look at the last forty years of technology, and a venture fund up the road that has sort of famously tweaked its peers by asking the question, you know, “We were promised flying cars and all we got was 140 characters.” And we could look at various types of technology and say, “Well, who was coordinating, centralizing? Who was doing some of that top-level-down thinking about it? And where did that get us?” Perhaps it got us the cars that drove us all here, but it also got us, you know, a hell of a problem in global warming.
So where do you go from there? You’ve got these little biohacker clubs, DIY bio. How do you start to think about coordination across those disparate groups?
Endy: It’s a good question, I don’t know. But then the next part of the way of thinking is what is worth coordinating? And it might be relatively little. You know, best practices with regards to safety is probably worth coordinating. Development of a language is probably worth coordinating. We humans develop languages over and again. Like we’re using English today, we have languages for communicating between people, but there’s also languages for communicating between us and machines or systems. JAVA, C++ is a type of example of these languages. We will have a language or languages for communicating with biology. And one can ask, how should we coordinate their development? Don’t know the best way to do it. It could be that the best way to do it is to make a lot of money, figure out the business model, and fund it.
So when I talked with Jim Clark about this, you know, when he developed the software to run one of his sailboats, he originally developed it on Linux, and then he kept having the kernel basically squirrel out from under him, and he got fed up with that and he moved all his software over to Apple, which is based on some things related to GNU/Linux, but he liked the fact that he could depend on a platform that was coordinated through capital and other things.
So all I can represent is there’s probably some experiments that need to take place for us, and me personally. We’re working backwards from a very deep view. We think that languages that survive over time are under selection to be free to use. And so we believe the language for programming life—I’m speaking here for the Biobricks Foundation, which is my side project—needs to be free to use language. And we’re working backwards on that basis. And by a language for programming life, I mean genetically encoded functions that cause cells to do things—float, sink, smell, turn left, die, whatever.
Lash: I just want to take a second and drop in—no slides, no data, thank you very much—but one little tidbit I picked up, and you mentioned: Is the best way to think about these broad questions as sort of give people a bunch of money and say, “Go for it,” in a for-profit situation? The Woodrow Wilson Center, a think tank in DC, reported at the end of 2013 there were more than 500 entities worldwide engaged in synthetic biology. And worldwide the universities and academics still outweighed the for-profits, but just by a little. In the U.S. alone, the for-profits have already overtaken the academics. So that perhaps is already taking place, that shift is already taking place.
Endy: Our students have to do something.
I mean, the other thing that brings up is question of organizational ecology. So one of the lessons for biotechnology to take from other fields of technology and economy are “how do we chose to organize ourselves?” The Universal Serial Bus Consortium is fantastic for what it’s done. The Apache Software Foundation—Brian Behlendorf’s here today. Mozilla, Long Now¾all of these different ways of organizing people to do stuff. Sometimes it’s for profit, sometimes it’s public benefit, sometimes it’s a university, sometimes it’s a national laboratory. Biotechnology has been starved in its ecosystem, I think. It’s been surprising to me to not see more experimentation with just ways of organizing. And if in the democracy of biology movement you see the community laboratories, the Genspaces, the BioCurious, also represented here today. That’s fantastic. It’s an experiment of organization.
Lash: So just maybe a more technical question, going back to languages and interfacing with biology; back when I was covering tech, before the dotcom boom, it was people who had to build webpages had to use HTML. And people were arrested for using the blink tag and all that. It was ugly. And then things got, you know, the interface got more modular, and all of a sudden now we have a much more sophisticated, top-level way of interfacing with machines. Where are we on that scale with biology?
Endy: We’ve shown it’s not impossible to make true. So the technical concept is sometimes referred to as abstraction: Can you create a layer of functions from low-level machine code up to higher-level functions that we could understand? So if we started talking in, you know, TAATACGACCTCACTATAGGGAGA, you go, “What’s that?” Right? You can memorize sequences of DNA by having them be your login passwords.
Lash: Have you done that in fact?
Endy: Yes. That sequence is the consensus promoter for the T-sub mRNA polymerase, for initiating the reading out of DNA.
Lash: Now we all know how to get into his computer.
Endy: Yeah, don’t log into my account.
But I can only memorize so much, right? So the idea of abstraction is to begin to black box and it’s very controversial, it’s very esoteric. So very few people have opposed us on that within the technical community. And it was only last year that we showed that we could do it for the first time. So we had previously spent three years and 700 design attempts bashing DNA to make what an electrical engineer would recognize as an analog-to-digital converter. We can take a continuously varying gene-expression signal from low to high, and it’s a particular threshold, flip DNA, or not. And it took us a long time to get that set up. But once we got that set up, we could bundle it and reuse it.
And we reused it. The first time we made it we used it to store data in chromosomes. This took us three years. The second time we deployed it we used it to make amplifying logic, Boolean logic, and every DNA design worked the first time. It took us on the order of ten weeks to do that project; it took us twenty weeks to publish the paper. And that is the only time I’ve ever experienced abstraction working, that we could go from low-level sequence to components to boxes, analogue to digital converters, and then make a family of systems on top of that.
So I would simply represent we’ve shown, and others, in a few cases, that it’s not impossible to make biology much easier to engineer. And that’s significant because if you can’t reuse things, then there’s no point in coordinating, right? Well, that’s probably not true. But many of the things you’d like to coordinate are not practically useful.
Lash: Well reusing things, then, there is no Mozilla—correct me if I’m wrong—Linux, in the biotech, biology world, given the—I mean, patents are fierce on the tech side as well, but biotech has its own unique set of circumstances. Do you see that type of model, open-source model, porting over? Maybe that’s what BioBricks—
Endy: It’s inevitable, it’s totally inevitable. So thank goodness the Supreme Court made a particular decision last summer that natural sequences of DNA cannot be claimed. I’m for that. And we worked with Mark Fischer of Duane Morris, out of their Boston office. Mark was the person Richard Stallman worked with to develop the GNU Emacs license that later became the GPL, got connected up with him through Saul Griffith at a meeting Tim O’Reilly set up a while ago, like ten years ago. And Mark drafted the BioBrick public agreement for us. We could not use copyright as the legal tool for giving and getting property rights around bioparts, so we created a scaling asymmetric contract system, which is interesting if you’re a lawyer, but it’s otherwise just a webtool, if you’re a contributor or a user.
The challenge we’ve run into is a legacy of the 1980s—most inventors are not liberated to give things away. And by dumb luck I find myself at a university where Kathy Ku in the Tech Transfer office is amazing, and if you read the Stanford patent policy, paragraphs one and two are just like everywhere else. Paragraph one, “You work for Stanford, we own your inventions.” Paragraph two, “We’re generous. We’ll share in the royalties if we license it.” Paragraph three—totally atypical. One and two are like everywhere else. Paragraph three says, “If the inventor would like to give the invention away to the public domain, you’re free to do so.” Period. Well we drive a truck through that one for the things we want to contribute to the public domain.
Lash: So you perhaps are in a unique situation. But what about the rest of the biology—
Endy: We’ve got to fix it. We’ve got to fix it. Period. I mean, sometimes we patent. We patent when other people are filing claims and we don’t like the claims and we need to make sure that we can push back against it. But when we’re ahead, the logic family, for example, we slammed that in the public domain. The lesson from the Stanford patent policy is, it’s nice to have options. Sometimes you want strong property rights. Public domain is a type of property right. It’s a property right optimized for sharing. So it’s good to have options. A lot of work to do.
So jumping back to your question, the operating system metaphor in biotech, not well developed yet. The language metaphor, the dictionary, the grammar—what does that mean? What is it that we are trying to compose? Do people really need to work together yet? Most of the projects in biotechnology are sort of short declarative statements. “Make lots of farnesene,” or “make a therapy,” right? But when we try to do more sophisticated things, like count how many times some things divide, invade that cell type, do biosynthesis of a benzylisoquinoline alkaloid, then after a certain number of division events, autolyse in a way that doesn’t trigger an immune response, right? So you’ve got censors, logic, data, memory, computation, actuation. We’re going to have to figure out what those metaphors from other technologies mean in living matter and make them true. And BioBricks, if we’re successful, will be part of contributing to those platforms.
Lash: I think a lot of the general cloud of worry around what’s happening, or what will happen with biology, comes from the fact that a lot of it is discussed with engineering terms—transistors, circuits, you just ran through a whole list of them. Well let’s put, sort of, the worry aside for a second, but keep going on that. You’ve got a foot in both camps, so to speak, the engineering and the high tech and the biotech. Do you find that it’s harder for one side to speak the language of the other or vice versa? And how do you navigate that as you—
Endy: I think this meeting is awesome, I’ve not seen a collection recombined in this form.
I think it’s difficult for both camps or factions, if you will. It’s also important to recognize that there are fundamentally different interests that have selected for biasing of expertise. So most of biotech, whether we admit it or not, is dominated by a scientific and medical mindset. Which is wonderful and important, but is very different than sitting next to the head of Fairchild or Intel, who are just raised in a different intellectual tradition of making. And so it’s difficult on both sides.
That’s why I think we have to create a new type of person. They’ll still be human beings, but they’ll be wet-ware engineers, right? And, you know, I’m embarrassed to be at an institution that helped invent genetic engineering forty years ago, but it’s only been in the last few years that we had an undergraduate major in bioengineering. We never started a genetic engineering department at Stanford. So there’s a lot of back-filling to do. It’s coming from, you know, I’m coming from civil engineering, and you know I worked at Amtrak one summer, so I’m used to a mature field of engineering. And to move into the space of living matter and want to engineer that, and just sort of, well, where’s all the infrastructure? It doesn’t exist yet.
Lash: But, to sort of stay on a top level, culturally it’s been very interesting to watch. I guess when genetic engineering began that was a real societal fear. And here we are thirty years later and roughly a fifth of the drugs worldwide are in fact recombinant proteins. And no one seems to be jumping in front of the Genentech buses to protest, but they’re jumping in front of the Google buses, etcetera, to protest.
As we move into the tech folks, the Silicon Valley folks, working in healthcare and biology, do you worry that there is a gulf that needs to be bridged, to understand what goes into healthcare, what goes into biology? And some of the concerns, say privacy, for example?
Endy: It’s a good question and I defer it to the subsequent speakers. And maybe we return to it.
I think, taking that question and returning to the previous one, one of the challenges I’ve encountered, and Nancy Kelley’s here, she’ll probably speak to this as well, is how people in other sectors of technology simply don’t know very much about biology and biology’s economic impact. And even though genetic engineering is relatively young, it taps into the reality that biology is very old. And so we can, as we apply our opposable thumbs to engineer living systems, immediately tap into huge manufacturing capacities.
Rob Carlson and a few others have tried to quantify the so-called bioeconomy, meaning, in their language, what is the dollar revenue, on an annual basis, domestically, for products manufactured using genetic engineering? The number, they estimate, is about $300 gigabucks, $300 billion, domestic. And drugs, actually, only $80 billion. Food is $100 billion and ‘stuff’ is $120 gigabucks a year—
Lash: Shampoo or what?
Endy: Better soap. Yeah, like Genencor, not Genentech, but Genencor making industrial enzymes and whatnot on Page Mill Road. That’s a lot of money.
And at Stanford, the way this manifests is we started this new bioengineering department and it’s amazing, we get to hire twenty-four faculty from scratch. It’s not enough! We probably need to double in size. Now when we talk about that on campus, the senior faculty go, “You could not possibly get as big as electrical engineering until you have as much impact as electrical engineering has had on Silicon Valley. The department has a whole element from the periodic table naming the region.” But what’s not appreciated is biotechnology is actually bigger now. We forgot that or we just missed that.
Oh by the way, having said that, I think this is, to use a simplistic metaphor, the biotech that exists right now is sort of the snowflake on the tip of the iceberg. There’s so many weird, interesting laterals. There’s so much more to make. How much of biotechnology have we imagined?
People gave some of the principals on these projects a hard time, but the two papers I really loved from last year were the ‘storing data in DNA’ papers, George Church’s book and then some Shakespeare. All the DNA was made in Santa Clara at Agilent, for both projects, on the inkjet print heads used to dispense the DNA. And maybe I thought it was a stunt initially, until the university librarian came to visit, and said, “We want to do this! We want to do this!” “You don’t know how unrealistic this is. This is $10,000 a megabyte to store and it has one month latency.” And he said, “You don’t know anything about the library.” That’s true. I actually have never been to the Stanford library. So he goes, “I’m part of a project that’s trying to secure a representation of civilization for 500 years”—not as long as Stuart, but 500 years—”and we have a problem with digital data storage because someone always shows up every few years and says, ‘Use our new format.’ And we don’t know how to forecast how much it will cost us to maintain the expertise to use the readers and writers of that format for 500 years. We can’t bound our budget projections.” So we rely on paper and we spend $100 million dollars to build buildings to store 60 million books. What if we could store it in that material that’s relevant to humans for as long as humans are relevant? And that’s on a path of cost down for read and write? We’re in. And we can afford one month latency, no problem.
So who would have thought? Who would have thought? I’m kind of shocked more companies haven’t shown up on just that little thing. It’s not going to be for our notes in classes and talks, but for archival data storage it’s already great. So I think there’s a lot more laterals.
Lash: So which part of—would you store it in your arm or your leg or?
Endy: I was talking to Kevin Kelly, he wanted to store his library in his gut. In the microbes in his gut.
Lash: But they’re always shuffling, so you get sort of a William S. Borough’s type of effect.
Endy: It could be. It could be “Naked Lunch.”
Lash: Literally, in your gut. So we’ve only got a few more minutes. We’ve sort of nonlinearly ranged across a lot of things. Is there anything you guys want to hear? Go ahead, raise your hands, shout something out. Yes.
Audience: What was the second paper that you liked?
Lash: I’ll say it, “what was the second paper?” You mentioned one paper, which was storing data in DNA.
Endy: Oh there were two papers on the same topic, competing with each other.
Man: So contracts changed, manufacturing’s changed, how does risk management change?
Lash: How does risk management change when contracts and other things are changing?
Endy: Yeah, and that’s part of a bigger question that’s sort of how do we represent what’s happening and figure out whether or not what’s happening is good or bad? And that won’t be black or white.
So, interesting conversation with DuPont, for example, trying to figure out how to scale up manufacturing with biology. They have a facility in Tennessee, for example, making a significant amount of biomaterials. And they represent that as being net-good, net-win in terms of resource loads on carbon and whatnot. It’s not all recognized how to scale the doing of such assessments. When you come to something like risk, what are the variables in the equation? Do you want to include biodiversity? Do you want to include land use? Do you want to include—well what should be in there? And how do we enable people to argue about that and not reach a decision, but continue to argue about it in a way that’s useful?
Now if I shift gears and talk about technology linked to risk, I think there’s a huge opportunity, both public and private market, associated with applying tools to de-risk aspects of biotechnology. As a specific example, the natural genetic code has been optimized by evolution to support evolution, or evolvability. Could we engineer a genetic code that an engineer might recognize as fail-fast. Fail-fast means if you’re launching a rocket and it goes off course, before it lands on the neighboring town you blow it up. So could you have a genetic code where every point mutation is deleterious? Nature would never select for such a genetic code. It would go extinct. Well I can fantasize about one. It’s not going to be a three-base code, it’s going to be a four-base code. I’m going to need 256 options. I’m not going to expand the number of amino acids, I’m going to keep that at twenty, and that four-dimensional space allows me to surround every coding codon with something that’s empty—weak, negative selection. Well that would be a heroic project, involving the chemists and the geneticists and everybody. And maybe it’s impossible. But you could start to create totally new, best-available public platforms to change the risk profiles. I think we need ways of sustaining conversations.
Another interesting lateral, by the way, returning to risk, is coming from the design community. You see this a little bit in Autodesk. They’re really a company that develops tools for enabling people to design systems, but also from the art community. And the professional design community—IDEO, the Royal College of Art in London. And what they’re really teaching me is that people—if we all went to a painting I would not insist that each of us have the same experience or opinion of the painting. If we all are presented with a biological technology artifact, I should also probably not insist that we all have the same experience, right? Yet somehow our conversations have become very one-dimensional when we encounter the artifacts we’re making. And how to bring that into risk I don’t know, but it is a good question.
Quinoñez: My name’s Carlo Quinoñez, I’m currently at Autodesk, and a big believer in the power of synth-bio and making novel genetic designs. And I really admire the making everything free, the basic designs. But in order to make those real, it requires hardware, right, to actually create those organisms, and then also to maybe sustain them and let them do their thing. So any thoughts on addressing the challenges of the high cost of proprietary hardware that’s necessary to complement the genetics?
Endy: Yeah, I think getting the prototyping at the wet-ware level by itself is one part of a much more complicated, expensive, trying process, whether it’s biomanufacturing or even worse in therapeutics. I don’t know. I do know that when you’re solving a complicated problem it’s good to separate it into simpler problems and take them one at a time. I think there are interesting models playing out right now at the start-up level, coming out of academic groups in synthetic biology that are commercializing. And the models being explored are sort of “Where in the design-build-test-bring-to-market cycle do you establish transaction boundaries for business-to-business relationships?” So, you know, Ginkgo Bioworks in Boston represents that the organism is the product. Other companies might represent that DNA, synthetic DNA, is the product. Well, which is going to thrive? When somebody places an order at a DNA synthesis company, do they really want a specific sequence of DNA? Or do they want a function that actually works? When you have an organism that’s been shown to prototype at bench scale, is that what you really want? Or do you want something that succeeds in process scale-up? I think there’s a lot of need to experiment with those process boundaries, those transaction boundaries, to figure out how to do better.
It’s a point where I’d like to imagine the semiconductor industry, for example, could come in and give us some perspective, you know, the MoSys service, Taiwan Semiconductor, fabless design. Could we recombine and leverage some of those lessons in a way that’s not totally naïve and that we don’t have to stumble through expensively to recapitulate.
Lash: I wanted to quickly ask about, the programming cells to store¾DNA to store data might sound totally far out. But we’re already—just kind of a point of clarification: a Nobel Prize was awarded a couple of years ago to Shinya Yamanaka and I think a couple others, in reprogramming—we are reprogramming cells with growth factors. And you’re talking about building upon the works of others to do—I’m curious if you actually are using some of that work to inform what you’re doing in terms of reprogramming cells as storage.
Endy: So when you have protein factors like that that cause differentiation of native cells, those basically become actuators for us. Those would be the things we control. We’re upstream of that with our genetic computing, so you could turn things on at particular places or particular points in time. What we’ve seen with the genetic data storage and logic systems we shipped over the last two years thus far, is people have moved them into medical settings, mostly with the microbiome, so organisms that colonize the gut. And you need to figure out “Is fructose present along with a signal from the immune system?” If so, flip the DNA and turn something on or record that data, so you can see what’s happening. So right now we’re developing basically diagnostic tools for discovering what’s happening in places where silicon doesn’t work. And those probably over a few years move into control systems for differentiation of stem cell fate and what have you.
Lash: I think we’re getting the high sign here. That’s it? Okay, one more question, sure.
Rote: My name’s Kurt Rote. I work with Western Oncolytics, which is a topic mentioned in the previous conversation, but this is a different question. So what do you think or what do you hear about in terms of engineering human intelligence? So right now we’ve already identified genes that affect human intelligence, and we already have the capability in in vitro fertilization to select fertilized embryos that may or may not have the selected genes. And so as of today, a couple could fertilize several embryos and increase the odds that their kids are a little bit smarter. And so that exists now and that’ll only ever increase. And if we start thinking about engineering human intelligence, then this starts to grow exponentially. The next generation then becomes much brighter, they can increase those tools, etcetera. And that would potentially have the potential to trump all other technologies or fields, whether it’s biology, computing, or anything.
Endy: So I like how Cindy Kenyon thinks about genetic screens for aging in worms and what the potential is. If you could imagine the last common ancestor between humans and worms and having a conversation at that point in life’s history, could you imagine an organism that lives 2,000 times longer? It would be impossible, but here we are. So what’s the potential for longer life?
When I think about intelligence specifically, and you hit me with that question on the fly, I want to talk to a physicist and ask them about heat transfer. The power required to operate the brain and the surface area needed for heat exchange and at what point would you start to melt the brain? So I don’t know the answer to that, but I’d come at that from a physics perspective as well as genetics. Thank you for the question.