Scroll Top

Richard Fox: Scaling Genome Editing To Drive The Industrial Bio-Economy


This week Harry speaks with Richard Fox, a computational biologist whose work at two life sciences startups, Inscripta and Infinome, is helping to automate and vastly scale up the process of engineering an organism’s genome to evoke new functions or uncover important genetic pathways.

With the discovery of the genetic scissors known as CRISPR-Cas9 in 2012, biologists gained the ability to make precise cuts in the genes of almost any organism. For genetic engineers, what used to be a slow, labor-intensive, manual process was suddenly easy. It was like jumping from a medieval monastery where all the monks write their manuscripts longhand into a world where everyone has a word processor on their desktop. But the first generation of CRISPR technology was still pretty limited. To continue with the word processing metaphor: you could use CRISPR to change individual letters in a text, but you couldn’t use it to modify entire words, sentences, or paragraphs.

At Inscripta, Fox helped to turn CRISPR into a fully-featured editing program. The company sells an automated device that can take bacteria or yeast cells and make thousands of programmed edits to different parts of their genomes in parallel. For researchers, a tool like that can vastly speed up the process of figuring out the relationship between an organism’s genotype and its phenotype. And that can help bioengineers create useful new strains of microorganisms—or uncover the genetic pathways that lead to disease in higher organisms like plants and humans.

And now Fox has left Inscripta to start a new synthetic biology company called Infinome. It’s a service provider that works with customers to design new types of organisms through directed evolution. The idea is to take Inscripta’s technology and add the power of data science and machine learning to speed up what Fox calls the “design, built, test, learn” cycle to create better custom organisms faster. The implications are mind-boggling—but in this episode, Fox walks through the ideas step by step.

Please rate and review The Harry Glorikian Show on Apple PodcastsHere’s how to do that from an iPhone, iPad, or iPod touch:

1. Open the Podcasts app on your iPhone, iPad, or Mac.

2. Navigate to The Harry Glorikian Show podcast. You can find it by searching for it or selecting it from your library. Just note that you’ll have to go to the series page which shows all the episodes, not just the page for a single episode.

3. Scroll down to find the subhead titled “Ratings & Reviews.”

4. Under one of the highlighted reviews, select “Write a Review.”

5. Next, select a star rating at the top — you have the option of choosing between one and five stars.

6. Using the text box at the top, write a title for your review. Then, in the lower text box, write your review. Your review can be up to 300 words long.

7. Once you’ve finished, select “Send” or “Save” in the top-right corner.

8. If you’ve never left a podcast review before, enter a nickname. Your nickname will be displayed next to any reviews you leave from here on out.

9. After selecting a nickname, tap OK. Your review may not be immediately visible.

That’s it! Thanks so much.


With the discovery of the genetic scissors known as CRISPR-Cas9 in 2012, biologists gained the ability to make precise cuts in the genes of almost any organism. For genetic engineers, what used to be a slow, labor-intensive, manual process was suddenly easy. It was like jumping from a medieval monastery where all the monks write their manuscripts longhand into a world where everyone has a word processor on their desktop.  But the first generation of CRISPR technology was still pretty limited. To continue with the word processing metaphor: you could use CRISPR to change individual letters in a text, but you couldn’t use it to modify entire words, sentences, or paragraphs.

My guest this week is the computational biologist Richard Fox, and he spent years working at a company called Inscripta that’s working to turn CRISPR into a fully featured editing program. Inscripta sells an automated device that can take bacteria or yeast cells and make thousands of programmed edits to different parts of their genomes in parallel. For researchers, a tool like that can vastly speed up the process of figuring out the relationship between an organism’s genotype and its phenotype. And that can help bioengineers create useful new strains of microorganisms…—or uncover the genetic pathways that lead to disease in higher organisms like plants and humans.

And now Fox has left Inscripta to start a new synthetic biology company called Infinome. It’s a service provider that works with customers to design new types of organisms through directed evolution. The idea is to take Inscripta’s technology and add the power of data science and machine learning to speed up what Fox calls the “design, built, test, learn” cycle to create better custom organisms faster. The implications are enormous, I’d even say mind-boggling. But in our recent conversation Richard took the time to walk me through the idea step by step. So let’s get straight to it.

Harry Glorikian: Richard, welcome to the show.

Richard Fox: Thanks Harry. It’s great to be here.

Harry Glorikian: Richard, I was putting together my notes, like on all the different things you’ve done and I’m like, Oh my God. I feel like I haven’t done anything with my life relative to what you’ve like accomplished. I mean, you started out as a nuclear engineer, but then you make this complete  turn into biological world. I’m making that assumption. I think you said somewhere, you read a book I can’t remember which book it was, that totally like flipped you into that direction. And then it was bioinformatics protein engineering. And then now gene editing. How, what, tell me a little bit about that.

Richard Fox: How did I get there? Yeah, no, that’s that’s right. It was a meandering path, especially early on, but then the last I’d say couple of decades has been pretty consistently in the field of biotechnology, especially protein engineering. And now metabolic construct engineering. We’ll talk a good bit,  I’m sure today. Yeah, I guess, sort of to rewind, you’re right. I did study nuclear engineering in college and I was working for the US Navy actually as an analyst, civilian. And I was on a ship actually out in the middle of the Pacific Ocean. And I had just been sort of spending my time reading a book called The Selfish Gene by Richard Dawkins. And it completely transformed the way I thought about the world, my place in the world, how evolution worked, and I was completely smitten with the concept of evolution and what it could do. The complexity that it could craft, that nature has done over billions of years. And from that moment on, I had a deep interest in evolutionary biology and the principles of that really elegant algorithm to optimize exceedingly complex systems.

So it wasn’t long after that, that I found myself ultimately, working for a biotechnology company. To be able to practice some of the principles  of evolution, although at a much smaller scale, at least.

Harry Glorikian: Yeah. It’s funny when you said the elegant algorithm and I’m like, wow. I wonder if there’s gotta be a lot of them, right, if you think about evolutionary biology. But I think the company you’re talking about is Codexis, was the one that you went to.

Richard Fox: Yup. That’s right.

Harry Glorikian: You worked on protein engineering, drug design, but relying on bioinformatics, statistical analysis, machine learning, evolutionary programming, sort of packing all those things together. Like, how did you bootstrap yourself into a position where you understood like all of these different components?

Richard Fox: That’s a great question. I think it’s like a lot of folks who take a keen interest in something. My career has pretty much been dominated by being interested in things and being passionate and being curious. And so those led to all of the experiences that I’ve had really that’s, that’s the short answer. It, it was driven by this interest in biology, but because I had in my studies in nuclear engineering that we talked about earlier, I had pretty much always worked on the computational side of things.

I was not good in the lab. I’ve never been good in the lab. I’m always amazed at what the scientists who can actually generate the data that I get to play with can do it’s stunning. But I get to sit and I get to play with that data. And I, for many, many years written software and algorithms to process that kind of data. And so that sort of was a natural, natural fit for me.

Harry Glorikian: So, just so people can sort of get an evolution of where you were, because we’re eventually going to get to where you are now, but, but so sort of what was the main focus of Codexis or the special sauce?

Richard Fox: Yeah. Great question. So actually Codexis was a spin-out from a company called Maxygen and Maxygen started if I remember correctly in the mid-90s with the invention of the technology a gentleman by the name of Ken Stemmer. He was really one of the pioneers in the field called directed evolution which ultimately went on to receive a Nobel prize that Frances Arnold won in 2018 for that field. Ken Stemmer was one of the great luminaries in the field early on. And he had developed this technology that would allow you to do evolution in vitro in the lab. Primarily around small or sequences evolving genes, proteins enzymes. And so that core technology was called DNA shuffling and it was very much all the main principles of evolution. So mutation, recombination, selection, all was being carried out in vitro. Very high throughput, very fast to evolve proteins and enzymes for different properties.

And so Maxygen, the founding company in the mid-90s got started and then in the early 2000s took that core technology and licensed it out to different subsidiaries. And the one that I was associated with, I started with Maxygen, but then I went with the Codexis  subsidiary, and they use that DNA shuffling technology to work with enzymes, primarily for pharmaceutical manufacturing processes.

Harry Glorikian: But, I mean, you invented, I think it was called proSAR while you were there, it was sort of to sort through protein mutation faster.  Should we have seen that as  foreshadowing to where you’ve sort of progressed to and where you are today?

Richard Fox: Yeah, I think so, actually it’s interesting before I had even joined Maxygen, so it turns out my wife was a scientist at Maxygen. When we first started dating, she told me about this really interesting company that she worked for and she described, it was evolution in the lab. And I was, of course already keenly interested in evolution as we talked about earlier.

And I think to this day, my wife wonders if I married her for directed evolution! Of course she’s a wonderful person, but I very clearly remember, or before I even came to Maxygen given the background I had in software and algorithms. I understood what they were doing in the lab. At least as much as my wife would describe it. And like most people with a background that I had, statistics and optimization, it’s sort of a natural, it’s sort of an obvious thing that you would want to do is given, given genotype and phenotype data. How can you search through that space more efficiently by trying to model the system?

This is something that statisticians have done, for decades. And it was just sort of being in the right place at the right time. So when I, when I then went to Maxygen I, and others, were very interested in applying these, these principles to the searching and seeking space.

Harry Glorikian: Yeah. And if you think about, computational capabilities, that whole space has changed, dramatically compared to what we could do in the ‘90s. But, and then you went on to, if I’ve got my history correctly, Inscripta. And tell the world a little bit about Inscripta, because I’m not sure how well it’s understood or how well the company is known.

Richard Fox: Yeah. So Inscripta is a life science tools company. So the easiest way to think about is they want to do for writing what Illumina has done for reading. So they want to be able to, at scale, intervene in the genome. Initially they’re working in microbes. Eventually they will be having a mammalian capabilities as well. But they want to be able to interrogate the genome at scale. They want to offer tools, benchtop tools, and reagents and software to be able to essentially automate and scale up as much as possible the editing process so that researchers can focus on their research goals and questions. Which is, if I intervene here and there and everywhere, as the case may be with these capabilities, and then being able to test a phenotype, what is the result of those interventions?

It’s hard to overstate how transformational that is, right? I mean, genome biology for years has more or less been dominated as an observational science with this ability to go and intervene at scale. You really, for the first time ever are turning it into an interventional science where you can really get at causality by making the changes in the genome rather than just reading them passively.

Harry Glorikian: I mean, if, if you can make the changes you want on human, the industrial application is unbelievable. I mean, the things that you could, right, design and then have that produce something that is for some particular downstream use, would be incredible now. But you guys weren’t using, I think you guys said you decided not to use the CAS9 approach. You use something called I think it was MAD7. So how, how did, what was, what was the motivation behind?

Richard Fox: Yeah, so it’s still straight up CRISPR. It’s just it’s with a different nuclease.  So early on Inscripta was looking at how to enable this high throughput, massively parallel editing capability. The inventor of the technology, his name is Andrew Garst, who was a co-founder of Incripta, and now he’s actually a co-founder of new company that that I started with him and two other gentlemen, and that core technology to be able to do high throughput CRISPR is based on the standard editing technology that centrally involves a nuclease. And Inscripta early on, was looking at the landscape around licensing and enabling researchers and because of some of the issues around licensing of the nucleases that were  out there, Inscripta made a concerted effort to go in and discover and develop a different nuclease. So it could still do the basic process of finding DNA and cutting it in the right place. But the advantage of using this other enzyme is this, that Inscripta offers it basically free to the world to use. So that they’re not encumbered by some of the more onerous licensing terms that are out there.

Harry Glorikian: So just so everybody kind of understands Inscripta, like, what was the process let’s say before Inscripta. And then now if you utilize something like the Inscripta platform.

Richard Fox: ah, great question. So CRISPR works. It’s amazing. And it well-deserved the Nobel prize in 2020. It is truly stunning its precision and its efficiency. But it’s still fairly low throughput. So if you want to go in and you want to make a change to a genome, you have to design your sequences so that they are targeted to the right location. Then the nuclease performs the cut and then there’s some repair process to usually insert the sequence that you want. And that can be done. by hand manually that design process can be scaled up obviously with computational tools, but you’d still be limited physically to doing only a small number of changes.

Just the molecular biology associated with bringing all the right reagents together is sort of can be a laborious process. If you’re making one or five or 10 changes, that’s not too bad. But if you want to make hundreds, thousands, tens of thousands, that’s a different proposition.

Harry Glorikian: Yeah. I was just thinking like, I’m just thinking, like even doing one or 10, like, and doing them. Right. And then now you’re talking about hundreds or thousands and doing them, it’s a completely different order. So if that’s what Inscripta does, then I almost, answered my own question of like…your data scientist group to plan out what you’re going to do has got to be, very good. I mean, your analytics capability. Is that what you spend the majority of your time working on and thinking about?

Richard Fox: For sure. When I was at Inscripta, that was the majority of what the team and I did, was to think about planning the experiments. And then ultimately when the data come back, you have, when you look at all the data coming back, you basically have millions of data points. When you multiply all of the sequencing data that comes back by the number of conditions and the number of edits that you have all across the system, you’re processing large sets of data to understand what each of these edits do.

Harry Glorikian: So,  if I, if I had to ask you like, so you’ve seen a lot, you’ve sort of made this evolution, and I want to get to Infinome in a moment here, but if you had to summarize sort of the impact of the computational methods that you’ve worked on on the biopharmaceutical industry, how would you sort of put that into context?

Richard Fox: It’s yeah, I mean, it’s hard to overstate the importance of computational tools. I mean, this, you couldn’t do much of this work without that, certainly on the informatics side of things, just managing the data. It’s not that sexy, but it’s of course critical. And then once you have all that data, actually turning it into meaningful insights. It’s profound. The algorithms for evolution do work though. And so one of the interesting things of the DNA shuffling technology that we talked about earlier worked without really a lot of informatics, you would basically apply, survival of the fittest to molecules and it would work and actually quite well, but it was ultimately a blind process at the end of the day.

And so to accelerate the fitness gain, you want to try and make use of that data to drive towards higher levels of performance in your system. And that really you can only do when you start interrogating what we call the genotype-phenotype map or relationship. And that’s allowed us to accelerate the process of evolution more than, than ever before.

Harry Glorikian: So that makes me ask the question of, is that what you sort of learned at Inscripta that guided you to start Infinome? Or was there other pieces of the puzzle that sort of the light bulb went on and you’re like, I need to go and start this next entity.

Richard Fox: Yeah, that’s a great question. So actually all of that sort of statistical modeling people call it machine learning now is, was done quite a while ago when I was back at Codexis. And to really understand the history, what happened was that we had developed a lot of these capabilities, but at the gene level engineering enzymes rapidly. So using statistical modeling, high throughput automation, software and information systems, and also a suite of sort of concepts about how to generate the data, plan, your experiments and best move quickly through the cycle, the design, build, test, learn cycle. All of that was very very well-developed. While I, and my colleagues we’re at Maxygen and Codexis  going back a decade or more. And so it was really around 2010, 2011, where that technology for doing gene based rapid evolution had evolved quite a bit. It still had room to grow and, and Codexis is is now arguably the state-of-the-art protein engineering company in the world. But what we were experiencing was a desire to move up to larger sequence spaces. So moving beyond just a single gene, we wanted to move to pathways and genomes because we believe the bio economy is in many kind of cases going to evolve, engineering, whole genomes.

Harry Glorikian: Right.

Richard Fox: And we were very excited at the prospects of being able to do this. And we had a strategy. We had a playbook, because we had developed it, to do single gene evolution or maybe a couple of genes at a time. Well, what we were missing, Harry, and this is where kind of to complete the circle with Inscripta comes in, is what we were missing for many years was the tools to be able to go in and make those changes.

Across the genome, as it happens, working with genes is fairly straightforward and has been for about 20 years. You can go in and diversify a gene very easily, very cost-effective. You can make all the single nucleotide or amino acid variants that you care to make. And then evaluate those though high throughput systems. You needed something like that, that ability to make those sequence changes, but at the pathway and genome level, and that’s what was missing for almost a decade.

We were waiting to apply this strategy, but we didn’t have the tool. And so that’s where Inscripta really came in was about three-ish years ago, I was very fortunate enough to get hooked up with the folks at the early stage Inscripta who were looking around at what to do with this massively parallel editing technology. And it was music to my ears and some of my colleagues is like, Oh, now finally, we can go after the whole genome, the way we’ve gone after genes.

Harry Glorikian: It’s sort of interesting that you can dial it up and then have these changes happen. I mean, if you, if I think back from where I started, like that was I don’t even know if it was a dream, it wasn’t even a concept when you think about it. And it it’s profound and scary sort of all at the same time, if, depending on who’s playing with it. But so now that brings me to Infinome right? So you went from, this protein engineering company that’s top in its field to Inscripta that seems like and correct me if I’m wrong, that’s working more on industrial applications of making changes to a bacterial genome or yeast or something like that. And now you’re at Infinome and okay. For everybody listening, including myself, what is, what is Infinome what’s it going to do? And how’s it going to change the world?

Richard Fox: Yeah. So Inscripta is amazing. The technology that they built and will be offering to the world is just transformative. Simply can’t overstate how powerful it is. And it’s more than just industrial applications, though. Plenty of biotechs, large and smaller, very excited about the technology. It also has a lot of application, basic science, antibiotic resistance, and all kinds of things that the academic community can dream up using this technology.

There’s lots of applications. So all that’s fantastic. What’s particularly a challenge on industrial side of the equation is, is that as amazing as the Inscripta platform is, it’s like any other technology stack. It’s one piece of the puzzle. It’s very important. It’s critical in many ways, but it’s not sufficient to do rapid genome engineering all by itself.

What you find, and it was, it’s also true, going back to the days at Maxygen and Codexis is that the core, DNA shuffling technology and then proSAR later and so forth, all really important pieces of the technology stack, what we found, because we were part of developing the whole ecosystem is that you needed everything else to work together almost seamlessly, to be able to run very quickly through the whole process. And so what Infinome is doing is it’s certainly going to use the Inscripta technology as a core part of its it stack. But then we bring together a host of other capabilities and experience or expertise to be able to run this in the synthetic biology world, the famous design build test learn cycle very efficiently, very cost-effectively.

Harry Glorikian: So is this a service? Cause Inscripta is a product per se, right, that might be sold to someone, but is, is Infinome more of a service of doing it because of all the different pieces that need to come together? Or can I buy this in a box?

Richard Fox: Yeah, no, it’s more of a it’s more than a service. I would say it’s a group of individuals with capabilities wet lab expertise, informatics expertise the know how to pull it all together. It’s definitely an execution team and a suite of capabilities. It’s not an off the shelf offering not by any means.

Harry Glorikian: So what do you say as like, assuming all of this comes together the right way? What, if you had to describe it to someone, what could you do? What would it be?

Richard Fox: Yeah, so it turns out there’s all kinds of opportunities in the bio economy that are just waiting for folks to go after, but they don’t have the capabilities to be able to execute on them. So the Inscripta technology is important, statistical analysis, high throughput, automation, all these things are important, but very few organizations have been able to pull them all together in a way that allows you to run very fast, very cost-effectively. And when you can bring that execution sort of an activation energy barrier, if you think about it, that way you bring that down. Now, a whole suite of bio-economy type applications are now on the table.

So certainly producing bioproducts, proteins, and small molecules that are high value or are commodity for that matter. They’re now all things that can, you can go after, because it doesn’t take, 20, 30 the people and 10 years anymore, like the way it used to, to engineer microbes. Very typical over the last 10 or 20 years for large engineering efforts that took many, many man years, potentially hundreds of man years and many tens of millions, if not hundreds of millions of dollars to generate these biological solutions. Now we’re able to do at a fraction of that sort of time and costs with the capabilities that, that Infinome will have.

Harry Glorikian: I mean, it sounds like though, I mean I always go through this debate of doing it for someone else versus doing it myself, sort of thing of you almost should do all the work yourself and produce the product yourself. That seems like it’s where it’s going to garner the largest value.

Richard Fox: Yeah. And actually that gets to Infinome’s business model, which is, we are indeed going down that road. So we are technologists. We love our technology, but at the end of the day, we, and I, I should have given you the background here. If it wasn’t already obvious, Inscripta was amazing. Great, fun, wonderful people. Some of the best colleagues I’ve ever had in my career. And yet where, what we found is that at the end of the day, we wanted to take this technology and apply it to actual applications. That’s what ultimately led to the formation of Infinome.

And so we ultimately had the idea that we wanted to build this technology stack to be able to apply to real applications. And as we looked around at how we wanted to build out Infinome it’s definitely a core part of our business. It’s sort of our reason for existence at one level, but we’re actually going to pursue some mix of both internal applications and working with partners, depending on how new, the opportunities that come into play.

Harry Glorikian: You know, I try to always in the show is get to like, that intersection of the biology and the data, right? The Inscripta platform sounds like it helps you efficiently apply the biology and know where to apply the biology based on the data that the informatics platforms that feed it. The question is now, in Infinome how are you looking at balancing those two pieces? Right. The data analytics at different points and, and getting the product you want in the end. Is it stringing together the right pieces of the puzzle to create something from end to end? I’m trying to wrap my head around these two concepts.

Richard Fox: Yeah. The data analytics, so that’s a really important question and piece to the, to the ecosystem. So as we’ve talked about before the ability to diversify sequences, whether it’s at the gene or the pathway of the genome is sort of step one. And especially in contexts where you’re making multiple changes, this is when the informatics becomes really important is when you have sequence variants where you’re making multiple changes, then there’s a deconvolution process to say, Oh, well, which interventions or combinations of the interventions are leading to the phenotype of interest. Right. And that’s where the statistical modeling machine learning really starts to be powerful. And so Infinome is in the process of generating lots of data, not with just single interventions, but multiple interventions.

And that deconvolution process will be, will be critical to sort of unmasking the genotype-phenotype relationship around the particular trait or phenotype of interest. This is definitely something that’s been done for many years at the gene level. It hasn’t really been done at the genome level, because again, we lacked the tools to make these things, these kinds of libraries, but now we have it.

And so now we’re off to the races again. So individual projects where you’re looking at, these relationships between genotype and phenotype certainly are amenable to this kind of statistical analysis. I think what’s really interesting is to think about down the road, how much of that landscape, that genotype-phenotype relationship, how generalizable is that? What are sort of the rules of thumb or guiding principles that you can apply across many projects? Maybe some of them are related. Maybe some are very different. What are kind of the patterns that over time with enough data, can you start to give yourself an advantage? When the next opportunity comes in, is there something that you’ve already learned from the data that you generated and the models that you’ve created, that you can apply to the future? This is the classic data network effect that we think biology has long promised to have. But I think because we haven’t had the tools to go in and actively intervene, we don’t really know yet what the boundaries of that, that possibility are.

Harry Glorikian: Yeah. I mean, it always seems like when we get to enough that there is a finite number of options that present themselves, depending on the model that you’re looking at. And I, of course, I mean maybe across different models, there may be that rule set may be different, but I think finding one and basing something on, which is why everybody seems to find one and then never move off of it because they spent so much time figuring it out. So, where’s the company right now in its process. ‘Cause I feel like it’s in, I want, I keep wanting to say stealth mode, but where are you in the growth phase or the gestational phase. Yeah.

Richard Fox: So we’re still early days. We we’re a few months into this. And so we were talking to lots of potential partners and investors, and we’re just about wrapping up our first round of funding. And we do have some partner projects that are spinning up as well as getting to work on our internal projects. So we’re going to be getting going here. We’ve been going in earnest, but we’ll be a little bit more public here very shortly about it.

Harry Glorikian: And if you had to like describe a perfect project, I’m sure that when everybody came together, they’re like, if we could do this, that would be right. As opposed to some, amorphous description of what it was. If you had to put it into brass tacks for people listening, what would you describe to someone as an ideal project from start to finish.

Richard Fox: Yeah, that’s a great question. I mean, it would involve at a high level, there’s the scientific, and then there’s also the business. And I can sort of speak to both aspects. So within business it’s not controversial, right? You want to go after high value products, right. Things where the economics around scaling the process. Are not so burdensome that,there’s already say commodity solutions out there. You’d like to go after things that maybe are at a smaller scale and sell at a higher, unit costs. Not to say that commodity solutions aren’t also our opportunities, aren’t also on the table. But that just comes down to techno-economic modeling and what, where are the opportunities where you can get into the market? And produce something better, faster, cheaper than something that’s already out there. So those are kind of typical sort of business considerations.

On the scientific side of things, there’s a lot of opportunities now with this technology that we’re developing that are putting things on the table that heretofore haven’t really been a possibility. So in particular, the whole space of natural products is a really exciting one. So it turns out that a lot of people produce natural products in sort of exotic organisms, because that’s where they’re initially discovered. And there’s large bias that there’s large gene clusters in these organisms and they just work.

And it’s for lots of folks, the perception is, is that, well, you do what you can do with what you have. That’s what you were given, what’s the old saying, you go to war with the army you have, not the army you want. And yeah. Part of it is, was based in some practical consideration around like, well, you spend all this time and effort to culture, these exotic organisms to do a lot of fermentation, process development and it’s working. But it’s not working well, but it’s enough to be economical. With today’s technology to be able to move large DNA sequences around recode them and optimize them for different organisms, and now with the ability to, once you have a microbe with say a heterologous pathway, maybe even really large ones from these other organisms, maybe 10, 20, 30 genes in them, now you can, with these high throughput, massively parallel gene editing capabilities and a suite of supporting pieces of the technology stack, now you can move through these pathways in genome sequence spaces much more rapidly than you ever could before.

So the barrier that was sort of there before, which is, well, even if I could move the pathway over, it’s still taking 10 years to get the bug to perform at the level that’s commercially viable right now, you can see a path where if I can move these pathways over in working much more engineerable systems, then I can get to that my commercial end point much, much faster than ever before. And this is not something that was possible before Inscripta and the Infinome technology platforms.

Harry Glorikian: Yeah. I can tell you, like, I mean, I remember we’d be working on a particular pathway and then, okay, we think we got it working, but let’s see how it goes. And then you’d have to wait weeks to get some sort of result. And then it’s not as efficient as we wanted. Let’s go back to the drawing board. And it would take forever for that loop to keep going back and forth until you, and I still say, hopefully, get to the result you wanted to, because there was no guarantee that you were going to tweak it to get it to do what you wanted it to do. Very painful process. Yeah. Yeah, it is. Because every time you feel like you’ve., every scientist will tell you I got it. I figured it out. I think we got it. I think we got it to do what we want it to do.

So if you took sort of… just so people listening can get sort of the timeframes because I’m, I’m big on this. The difference between evolution and revolution is time. If you wait long enough the change will happen, but right now, what I see is technology accelerating things and, and the timescales are being collapsed  at much tighter timelines. If you had to talk about where we were sort of in genome editing and then put that into a timescale and talk about where we are now, how would you.

Richard Fox: Yeah, it’s a great question. So the core editing technology that Inscripta has developed is orders of magnitude more efficient. I mean, there’s, there’s things you can do with the Inscripta platform that you, you just would never consider doing by hand, to make 10,000 edits or more across the genome, which try to do that by hand, it would just be, it wouldn’t be feasible economically or manpower wise.

So that ability to do massively parallel editing is sort of without a comparison. You just simply would try fewer things. And it would probably take you even more people with existing molecular biology techniques. So that’s already one, like, several order of magnitude level of efficiency. And then as we talked about earlier, as amazing as that is, even that’s not sufficient, right? Because now you have all these variants

Harry Glorikian: Right.

Richard Fox: Now you have to be very efficient in testing them. And it turns out that that’s also a bottleneck. And so even with some of the best folks out there today practicing genome engineering, you still find that the teams are fairly large and relatively slow when it comes to processing these variants.

So, and this one’s interesting because it’s not that the technologies and the strategies don’t exist to do it. It’s just very rare to find the, sort of the folks who can bring it all together with the right information systems. Lean smart automation. So to give you some numbers, for example, and I’ll actually, I’ll go back to sort of enzyme engineering back, 15, 20 years ago, teams would be 10, 15 people. You would do one round of evolution, maybe every couple months, and after a couple of years or more, you get to your end point. Now state-of-the-art enzyme engineering teams are much smaller, two to four people, one round every two weeks, maybe a month in the slower projects. And so you’re already seeing multiple factors of speed-up in the enzyme world.

It’s that same sort of step up that we’re looking to do with pathways and genomes, so much smaller teams, maybe a quarter of the size or smaller. with much more diversity going into the pipeline, thousands, tens of thousands of things that you’re testing. So when you multiply that out on a number of things, tested per unit person, it’s maybe three orders of magnitude more efficient.

Harry Glorikian: And so if, if you said, so now I need a quarter of the people or a third of the people let’s say. I’m able to do more. What is driving that? Is it, is it the data science side of it? I mean, I feel like a lot of the biology has been there already, but is it in the industrialization of the biology plus the data science?


Richard Fox: It’s both. I mean, it’s definitely, as we talk, you couldn’t do this before, these high-throughput, before this massively parallel editing technology was developed, you just simply couldn’t. So that was a key piece that sort of opened up the floodgates. But now it’s, a lot of it is managing what you create both physically and the downstream tests, software and information systems to manage all the data and quickly and intelligently getting to the next round of prescribed experiments that you want to do without all those pieces. You simply would be sort of hobbled in the overall sort of cycle time and how much functional gain or leaps in fitness you can affect at each, each round.

Harry Glorikian: Okay. And then it’s tweaking at every single one of those stages to make each one better or more efficient.

Richard Fox: Yes, exactly. Yup. And sort of a key thing, it’s sort of an obvious point, Harry, but it’s, it’s interesting after all these years that it’s not widely appreciated is the following, which is in every step of design build, test, learn, there’s—to steal the term from electrical engineering—there’s an impedance mismatch, right?

So between build and test, for example, historically, there can be widely divergent throughputs for build or test. Sometimes you can only build a few things. And you’ve got a really high-throughput test. Or vice versa. And so what we’ve seen, what we personally experienced and been involved in innovating around is to minimize as much as possible that impedance mismatch between every step of design, build,, test learn. You can make orders of magnitude improvement if you pay attention to those mismatches.

Harry Glorikian: Yes. And I always think about it as whack-a-mole. I fixed, I, I make one part of it better, the bottleneck just moves, right. It just moves where it is. And I don’t know if I ever get to the whole thing is just moving at the pace I want it to, because ultimately there’s only so many things you can pay attention to at the same time.

So, so you’re telling me that basically what might take me three or four years to do by historical or old methods now might take me. Six months to a year.

Richard Fox: Yes, that’s right. With, at a fraction of the resources as well. So it’s not just how long it takes. It’s integrating that resource burn over that period of time. Possibly, a factor of three to five, perhaps even more integrated over a longer period of time. We’re looking at much smaller teams, much more efficient use of resources. Getting to the end point much more quickly.

Harry Glorikian: So who is this disruptive to assuming we can do all of this, right? Who is this disruptive to out there?

Richard Fox: There are many sources of disruption. I guess one would be, depending on what you’re going after, for products that are based on saythe petroleum industry. If you could move those into bioproduction processes and replace those other sort of conventional sources of production, then it would be, those sort of old style of petroleum-based producers.

So they would be potentially disrupted by this. The way I like to think about it is, is that, it’s a big world and sometimes people ask, well, what is Infinome’s long-term plan to do. And while we definitely want to create products and be successful, our view is that it’s a big world out there and that there’s so many opportunities to go after.

We’re excited, just sort of as scientists and, members of the human race on planet earth. We are very excited that long-term, these kinds of approaches will find wider adoption now that the tools are coming online. And if we can help be a part of sort of blazing the trail there’s a part of us that would be very fulfilled and satisfied if we can see this technology getting used in other, other areas as well.

Long-term, if we can help be a part of that process, either actively or passively, it’s up for debate and it’s one of the business models we’re considering, which is, as we get better and better at this and execute on multiple projects, both internal and external, eventually, if we can help the rest of the world in some way as a template, possibly, licensing technology expertise and so forth.

Because as I say, there’s no way that Infinome, even if we became, a huge company like Cargill or DSM or ADM in large manufacturing. Even for them, the world’s a big place, right? So we’re very interested in pushing the envelope, being successful on what we go after and then ultimately hoping that and being a part of, creating the ecosystem that the rest of the world can also use to go after the countless bio products that are going to be developed over the next 20, 30 years.

Harry Glorikian: And it sounds like over time as you’re accumulating the data and understanding, I make this change in these, these are the implications and this is what happens downstream. I mean, at some point it becomes much, much more data science than just, what chemistry, at some point, if you’re focused in a couple of very discrete areas.

Richard Fox: Yeah. I think that’s right, Harry. And that gets to this really interesting unknown at this point of how much can you generalize the process and the information that the models that you’re learning? How generalizable are those to other parts of the genome.

So I’ve already mentioned this sort of sequence-function landscape several times. It’s a concept that’s been around for almost a hundred years now. If you think of you genotype as latitude and longitude, and elevation as phenotype, if you think of nature, having developed lots of mountains and hills across this, very high dimensional sequence function landscape. A really interesting question is, if I’m climbing up this mountain for product A, if I go after product A’ and it’s similar to A, arguably I can use some of the information or a lot of the information that I’ve developed already around product A to extrapolate to A’.

I think what we don’t know yet is, if you go for product B and it’s near A, but it’s somewhat distant, how much can you extrapolate from what you learned about A and A’ over B? And this gets to, is it really in the cards that you can create a global sequence-function landscape for all possible traits and phenotypes? That is a very tall order. I don’t imagine that’s going to happen in my lifetime.

Harry Glorikian: I agree with that

Richard Fox: The models for navigating these spaces, I think definitely are generalizable, but then it gets down to how close do the landscapes need to be similar to each other for you to leverage what you’ve already sort of learned about them.

Harry Glorikian: But at some point, right, you get to know A well enough that there is, there’s an informatics approach to it. And that it’s going to work because you’ve worked with it so much. And then you get to know A’. Right? I, I understand the generalizable, which would be awesome. But even as you’re moving down, some of these product areas, somebody comes to you and say, can you make that tweak for me?

Richard Fox: Yes.

Harry Glorikian: It becomes a lot easier to make the tweak than where, when you first started trying to understand A well enough.

Richard Fox: That’s right. That’s spot on, Harry. That’s exactly right. And so if you’re working in related product classes then there’s definitely huge value built up over, proprietary, data sets and models generated. You can definitely leverage that move much faster. than if you were starting from scratch, for sure. Yeah.

Harry Glorikian: Yeah. I mean, it’s funny, right? I always used to say to them, I ran a consulting firm for a while, strategy consulting, and I’d be like, the first customer that comes by then, we’re going to do our best, right? The fifth customer, man, they got such a good, insight, an answer, because there were five that we learned from, and we knew exactly what was going to happen. But, and I look at this the same way, but, but with more solidified data pathways, understanding what changes cause what downstream. And now someone says, well, can you make this slight tweak for me? It’s not starting from scratch. There’s an informatics backend that sort of, you can dial up and get what you want. And so the timescale of being able to do it would also be less. It will also shrink.

Richard Fox: Yeah, that’s right. That’s right.

Harry Glorikian: Well, all this sounds super exciting and super scary all at the same time. Right? Cause I can think of all the great stuff that can be done, but then I can also think of like, the easier and easier this technology gets, the more you worry about who’s doing that work.

Richard Fox: Yeah, I, that’s a good question. And that one, so just so Inscripta takes that [seriously] along with a lot of people who work in this business. The gene synthesis providers have faced this for many years and they have taken that very seriously. So they, they screen for nefarious sequences or uses that could potentially be problematic. Inscripta is the same way. You can’t just order up whatever you want and create new pathogens. There are pretty strong restrictions against doing that. So it’ll be interesting to see, going forward, how companies like Inscripta and others will continue to stay ahead of this. I think it’s very important for them to take an active role in this and not because the alternative is, is that the government would step in and legislate and create a lot of bureaucracy and slow down the science.

And so I think the industry behooves them, all these tool providers and users, it behooves everyone to try to do the right thing here. And so far, everything that we’re seeing from Inscripta and in other companies is that they are, and they are taking this seriously. And they’re putting methods in place to prevent uses that could be dangerous.

Harry Glorikian: Yeah, no, that’s good. But it’s interesting because this, this whole area that you and I are talking about, the implications are profound and I’m not sure everybody fully, I’m not sure that most people appreciate how quickly things have moved compared to where they were, I don’t know, I want to say 10 years ago. I mean, 10 years ago, it feels like a lifetime, when you look at the level of change that’s happened, across the board.

Richard Fox: Yeah. It really is stunning. I mean, I, the first I remember being in Inscripta and seeing the first real large-scale experiments, that I was involved with at least. And seeing that come out and seeing that we were literally editing, five, ten thousand different genomes with things that we precisely designed and wanted to have integrated into the genome.

I couldn’t believe that I was really looking at the data that was really corresponding to reality out there and that we had created. 10,000 new organisms. I mean, in a precise way, people have been doing random mutagenesis, but like in a directed, precise conscious way having that power. I’ll never to be able to describe it. It was, yeah, it was something as a computer guy I have long wanted, because I can sit and write out sequences, and I’d always wanted this ability to do this for genes and pathfways and genomes. And so to actually finally hold it. It was it was really special.

Harry Glorikian: it’s funny because I’ve always said over the years, like biology always, you can come up with a great thing. You can map it out, you can do all the work you want. It doesn’t mean that biology is going to participate willingly. Right. And now what you’re saying is, is we’re getting a whole lot better at figuring out how to get the, the software of biology to do what we want it to, or, or manipulate the hardware within biology. However you want to look at it, but to get it to do what we want it to do when we want it to do it.

Richard Fox: Yeah. I think that’s right. And actually, we didn’t really linger on this, I had talked a lot about my interest in evolution, but just to be very explicit about it, because it’s important: The reason why this technology is so important is because we don’t know the rules of biology.

If you knew the 10 or 20 changes that you needed to make, and you just went in and made them, and from first principles could design these biological systems, it would be wonderful. And there was a lot of interest in synthetic biology when it first started gaining currency as a term 10 or 15 years ago, that was the aspiration.

And that was certainly laudable, but it’s met with very limited success in the way that a mechanical and electrical engineer would think about engineering a system. This is just not in the cards for biology anytime soon. So being able to try lots of different things is critical to being able to get to your desired influence faster. This is something we’ve known from proteins for many years, and it’s always been true, of course, at the larger sequences of pathways and genomes as well.

Harry Glorikian: Yeah, I see it across, multiple areas,  materials, chemistry, there’s all sorts of areas where people now are applying, machine learning and AI. The properties that they’ve got from the chemicals that they’re working with and being able to just go through a giant sort of figure eight and just keep testing out until they figure out what gets this thing to get to where they want it to be and then being able to make it reproducibly.

Richard Fox: Yup. That’s right. Yeah. I mean, there’s a reason Frances Arnold won the Nobel prize in directed evolution and not a computational protein engineer. As amazing as the work they’ve done, it’s just, you can’t design a protein from first principles to get a 4,000-fold improvement for some property of interest. It’s just, it’s not possible. So you have to try many things and let nature tell you what works and what doesn’t. And it’s absolutely the same for pathways and genomes as well.

Harry Glorikian: Yeah, I guess just to summarize it here, though, what we’re saying is we’re going to start telling nature. What we want it to do and it’s going to do it for us.

Richard Fox: Yes, exactly. Maybe over time, as we’ve talked about, some of these patterns will become emergent, especially around A or A’. But, the full, the full truth behind nature will be, I think, hidden for the foreseeable future. So we’re going to have to rely on empiricism,

Harry Glorikian: I think to, yeah, I’m happy to take it one, one at a time, one step at a time is fine. You can still make a big difference in people’s lives in the environment and that’s what we’re in this business for. So, well, it was great to catch up with you. I do want to talk to you once things are up and running and hear how the dream is becoming, the fulfilled reality. But maybe we can stay in touch and, and, and touch base at that point.

Richard Fox: Yeah. Yeah, o, this has been great. I’d be really excited to share with you some of our early successes. Once, once we get going and you start to talk more about it.

Harry Glorikian: Excellent. Great talking to you.

Richard Fox: Great. Thanks, Harry.

Harry Glorikian:That’s it for this week’s show. We’ve made more than 50 episodes of MoneyBall Medicine, and you can find all of them at under the tab “Podcast.” You can follow me on Twitter at hglorikian. If you like the show, please do us a favor and leave a rating and review at Apple Podcasts. Thanks, and we’ll be back soon with our next interview.


Related Posts