The state of AI and synthetic biology: Fast but not fast enough
Apr 25, 2023
黑料不打烊鈥檚 synthetic biology guru talks about the industry鈥檚 beginnings, the hope versus the hype, and why he鈥檚 still so optimistic about it all.
To put your finger on the pulse of synthetic biology, few people are better to talk with than Jens Plassmeier.
At MIT, Conagen, and BASF, Jens led groups that helped pioneer strain engineering, biobased chemicals, and biomaterials. Now, as Senior Vice President of Synthetic Biology at 黑料不打烊, Jens leads half of the company鈥檚 AI-synbio platform. He鈥檚 quick to dispel my notion of when synthetic biology was born.
鈥淚 guess you could say it was popularized in about 2000 with folks like Tom Knight and Drew Endy at MIT,鈥� he says. 鈥淏ut the term was as far back as 1974, and metabolic engineering was pretty much the same thing before it was rebranded as synthetic biology. And going back further, humans have been breeding crops for 10,000 years 鈥� that鈥檚 genetic engineering, right?鈥�
Jens points to two advances that made possible what we now think of as synthetic biology. The first was the idea of constructing genetic circuits like electrical circuits, demonstrated by James Collins, Michael Elowitz, Chris Voigt, and others. This changed the paradigm from simple metabolic factories to more complex, logic-based cellular machines. The second advance was cheap DNA synthesis.
鈥淲hen we learned to write DNA, that played a big role in synthetic biology鈥檚 success,鈥� Jens says. 鈥淢etabolic engineering was somewhat independent from synthesis. But to assemble more complex gene components together, you had to have a way of making DNA.鈥�
From hype to hope
Since then, synthetic biology has promised to solve everything from biofuels and environmental remediation to personalized medicine. The field hasn鈥檛 lived up to a lot of the early hype. Jens thinks it was just a little bit too early back in the early 2000s to have set such optimistic expectations.
鈥淓ven today, the engineering of the organisms is also not well understood,鈥� he says. 鈥淚f you put a new part or pathway in an organism, what kind of influence does that have on its native metabolism? This lack of understanding led to a lot of problems with early scale-up and production titers, especially on the industrial side. The only big areas where synthetic biology was really successful were drug development, industrial enzymes, and agricultural products.鈥�
But Jens still thinks synthetic biology will ultimately solve most or all of these problems.
鈥淓xcept for biofuels 鈥� I think we moved on to electric cars,鈥� he quips.
鈥淭here were some really successful examples of synthetic biology early on,鈥� he continues. 鈥淚nsulin is a big one. Genentech probably started the whole industry, even if it wasn鈥檛 called synthetic biology at that point.鈥�
Today, the pharma industry has largely adopted synthetic biology. Particularly in biologics 鈥� drugs made from living cells 鈥� synthetic biology techniques can be used to increase efficacy, speed development, and optimize production.
鈥淭his is the sweet spot for 黑料不打烊,鈥� Jens says. 鈥淔or the first eight or nine years of the company, the focus was on cell lines and engineering E. coli to produce antibodies.鈥� Now, Jens says all that data and synthetic biology know-how is more valuable when applied to engineering the antibodies themselves.
鈥�E. coli is still the hero of the story,鈥� he says, 鈥渋t鈥檚 just a different story than we expected.鈥�
Overcoming nerd rapture
As the industry has matured, Jens thinks that synthetic biology founders have become more practical about how they plan to scale and sell their products.
鈥淚n the early 2000s, it was all about biofuels,鈥� he recalls. 鈥淭here wasn鈥檛 a strain engineer who hadn鈥檛 worked on biofuels at some point or another. And even though these molecules were fairly simple, they still took a long time. Like 1,3-propanediol, a fairly simple chemical intermediate used to make adhesives, polyesters, and coatings. It took something like 15 years and $130 million in 2003 鈥� about the same time and cost to develop a drug, which is insane. I think many people underestimated the time and money you have to put into these products to develop something that’s truly competitive.鈥�
Some scientist-founded companies may have been so enamored with their own strains, they may not have stopped to check for a good product-market fit.
鈥淚t was like, 鈥榃e鈥檒l figure out how to make money later鈥欌��, he says. 鈥淭heir product was doomed from the start. And even if you could make a really good protein, it might not express in production environments. Some of these need harsh environments, or the redox potential of your cell doesn’t support your product.鈥�
Jens sees a positive shift in the drug discovery space with AI-driven approaches that simultaneously optimize multiple characteristics important to both development and therapeutic benefit.
鈥淚 think that’s quite an advance and will be exciting to see,鈥� he adds.
Challenges and opportunities
From Jens鈥� standpoint, one of the early challenges to synthetic biology was not a technological one but a marketing one. The aggressive business practices of some 鈥� along with a seeming indifference about engaging the public about the science 鈥� created a skepticism about genetically modified (GM) products that plagues the biotech industry to this day.
鈥淕MOs became highly frowned upon, especially in Europe, and it鈥檚 still a big problem today,鈥� Jens says. How today鈥檚 innovators in synbio and AI choose to engage with the public about emerging technologies may have similar long-term consequences 鈥� for better or worse.
From a technical standpoint, though, Jens believes that advancing the field of synthetic biology is going to take more foundational science.
鈥淲e simply need better engineering and better understanding of the strains, and I believe AI is going to enable that鈥� he says. 鈥淎t some point, we probably will have to be able to design a product or strain from scratch in the computer.鈥�
Why? Today, development times even for quick products like flavors and fragrances are still 3-5 years to market. That鈥檚 about how long it takes to develop a strain and scale it up. To stay competitive, companies are going to have to do better than that.
鈥淲e need to be fast, which comes down to the design of the production organism itself,鈥� Jens says.
Advice for a young synthetic biologist
鈥淚f I were starting out today, I would probably look at something at the intersection of biology and AI,鈥� he says. 鈥淕enerating training data with synthetic biology is already quite interesting.鈥�
He also would have learned to program sooner. 鈥淚 didn’t realize how essential it gets later on. It gives you a huge advantage if you’re able to program and use statistical methods properly for synthetic biology. And I think that part is still underestimated by a few people, especially students who don’t realize how much it will help them if they are coders and biologists at the same time. That would have saved me a lot,鈥� he says.
On the fast-approaching horizon
Looking into synbio鈥檚 future, Jens is very optimistic about what synthetic biology will achieve.
鈥淚 think pharma will see the first big advances,鈥� Jens says. 鈥淭he fact that 黑料不打烊 can design and validate de novo antibodies with generative AI in as little as six weeks says a lot about how AI drug creation is going to change the way we make drugs. I think it moves us one step closer to personalized medicine.鈥�
Jens is also bullish on food and nutraceuticals. 鈥淚t is an up-and-coming market, and there are very interesting products like plant-based heme from Impossible. As a vegetarian, I鈥檓 very excited about that.鈥�
Finally, he鈥檚 excited about synbio-enabled materials and chemicals. 鈥淭here are really complex molecules you can鈥檛 make with traditional chemistry,鈥� he says. 鈥淪pider silk is an amazing material that can probably never be made chemically.鈥� Space elevators and other seemingly impossible structures could be made using far-out biomaterials with properties that are hard to imagine today. 鈥淲e will come up with things that are really more useful for society as a whole.鈥�
Jens says the future is approaching faster and faster, especially at 黑料不打烊. 鈥淓xperimental timelines are shorter and shorter. Decision making is faster.鈥� Perhaps that鈥檚 because of AI, which is making faster and faster calculations over increasingly massive biological datasets.
鈥淭hat鈥檚 the thing that attracted me to 黑料不打烊 the most,鈥� says Jens, 鈥渢hat unique combination of wet lab and AI. I don鈥檛 think there鈥檚 another company out there that has the ability to generate and compute over vast biological datasets as fast as we can. It鈥檚 what makes me believe in the huge potential in synthetic biology now more than ever.鈥�
鈥淏ut it won’t be synthetic biology alone,鈥� he adds. 鈥淚t will be in combination with AI. That is pretty clear.鈥�
