Founder here; working on software that exploits the full potential of engineered microbes; we're using them to make therapeutics, plastics, materials, etc. happy to answer any questions.
When I was studying chemical engineering in university, I tried to get faculty interested in engineering yogurt-producing bacteria (like L. bulgaricus) to produce Vitamin A (or an equivalent retinoid or carotenoid): a "golden yogurt" scheme like Vitamin A producing golden rice [1]. But, they weren't having any of it.
This could be useful for the 670,000 children who die [2] and 250,000 to 500,000 children who go blind from Vitamin A deficiency [unsourced].
The yogurt could be produced from dairy and plant stock (which is presumably easier for subsistence farmers to procure).
I had the good fortune of switching careers into software, so I don't have the wherewithal to do this myself.
If you happen to have time, I think you could produce pretty good results from engineering a Vitamin A producing strain.
Where do you guys stand on "open source" synthetic biology (like BioBricks or OpenWetWare)? Will we see any source from you guys, whether it be of the von Neumann or the Crick variety?
Completely understand your sentiment. I have been releasing software as open source for 10+ years -- all of my 7 years of program synthesis PhD sw is available as open source.
With 20n, we had to tread more carefully. As monocasa implied, open source here means either code, or dna data. While code stays on the computing platform, dna's platform is every living entity.
Lets assume you could predict precisely what a piece of DNA will do -- we are not there yet, but will be at some time in the future. Are you advocating that all of that should be freely available? There are lots of benefits to open source and we understand a lot of it in the context of code, but IMHO, we will need to discuss more before we fully comprehend what it means to open source dna.
There are tons of opportunities, e.g., enabling bioproduction, and hopefully the beneficial nature of that (cheaper/carbon negative/environmentally-friendly) will be the leading argument in the conversation than anything else.
Thanks for your comment. As I said, it's understandable for your company to keep its software proprietary, after all Google for example, while a supporter of and contributor to open source, does not open source critical components of its business either.
On the wetware side, there is of course a similar incentive to keep the genetic sequences that result from your work private, in addition to perhaps ethical concerns regarding the dissemination of arbitrary DNA.
As a fellow software guy considering going into drug development research though, I do wonder: Given your mentioned open source credentials, it's probably plausible that you are at least sceptical about software patents? If so, what is your opinion on patents on drug molecules and DNA sequences? I'm struggling a bit with this since the only way to a big exit seems to involve protecting any findings as IP (because of the costs of the regulatory process), but I'm not sure I'd want my work to be patent-walled.
Do you know what you would do if you are one day faced with the choice of whether or not to patent one of your molecules or sequences, knowing that not doing so would have significant adverse financial effects on your company?
(This all may sound somewhat critical, but I'm really just curious. My email is in my profile if you prefer to respond privately.)
Given that "reverse-engineering" DNA is trivial compared to "reverse-engineering" small molecules or reverse-engineering software, do you have an opinion on what constitutes non-"open source" DNA?
Just because Ken Dill declared victory on the protein folding problem [1] doesn't mean it's going to become trivial any time in the next few decades...
EDIT: after checking the paper, it looks like he qualified himself far better than I remembered. Maybe I was thinking of a different paper? In any case the lower-left part of the last page is a good partial list of reasons why reverse engineering DNA is not trivial.
What safeguards are you putting in place to not repeat the Klebsiella planticola debacle [0].
What is the possibility of an engineered organism escaping into the environment and over producing a beneficial substance, insulin, acetaminophen, Lysergic acid, etc.
Any plans to make this available to others who are looking for novel pathways for other (non-microbe) applications? Specifically there are a few chemicals we want to express in plants.
this sounds amazing! aside from the organics-only requirement (which makes sense to me), what other limitations are there on what you can make? or is everything basically open given a long enough search?
Many limitations! Constructing the designs (i.e., going from data->dna->cells) takes time and is expensive: we pay vendors for the construction. So we have to restrict our software to output designs that yield the most novel products, with the least change to the biology of the microbe.
Most of the time changes do nothing! But if our models are accurate enough, the engineering can be predictable. We are continuously improving our models -- from more data mining, and from feedback from experiments.
That said, what our algorithm predicts right now is still a huge space (5000-10000 products). That is the tree image in the article. For context, the chemical industry centers around 70,000 products.
Really cool. Question: Assuming that your algorithms will become known or can be replicated, how do you think about your defensibility? What is proprietary here that improves via network effects? How does the licensing model work and what are some comparable (old school/current) companies?
Defensibility for us: microbes we build, and the platform they form. Our software-driven engineering is significantly faster than the state-of-the-art: human's looking at metabolic maps. People will start building on top of our initial microbes and improving them. These initial microbes form the "platform" from which to enable more bioproduction apps.
> Defensibility: .... People will start building on top of our initial microbes and improving them. These initial microbes form the "platform" from which to enable more bioproduction apps.
That's not a very fair answer. I have often said pretty much that and was laughed out of the room. Granted, maybe it's all in the delivery.
Microbes can be easy to culture and store, so that doesn't really count as defensibility... and the software can't either, because nobody these days is really struggling to run algorithms over regulatory networks or KEGG or whatever...
There is already a lot of competition here (which is not necessarily a bad thing)...will be interested to see how 20n can differentiate and or add value upon what already exists.
Indeed! We love this community [1]. We started 20n with DARPA's help, in part to gather more steam around the accuracy of software predictions tools within synthetic biology. You seem to have worked on these tools, and so must know the pain of convincing wet-lab scientists to follow up on the sw predictions.
Paracetamol, both for academics and to pharma, was a non-biosynthesizable molecule. But once we had the prediction, we were able to go to the lab to construct the microbe pretty easily.
In the end, the tool is the start of the process. We are investing significant resources in constructing the microbes. Over the coarse of the next two year, hopefully you will find some novel molecules in the ones we move to bioproduction.
[1] Have to! We are part of it. :) We know most of them, and they probably know us.
Cool! I almost went into a lab studying polyketide synthase enzymes. How heavily do they feature in your pathways? Is their modularity overstated/understated?
(I fully intend to stalk you on google scholar after I get back to the things I really ought to be working on right now...)
Hah! Oh PKS-es! Megasynthases (PKS, NRPS, FAS) do not currently feature very heavily in our algorithms, but their modeling is certainly within our intended algorithmic expansion plans. As for whether they are understated/overstated, my cofounder can sit you down for hours and speak about them, but you have probably moved on!
ps: Googling me will lead you to program synthesis: programs that program programs :) which is another curiosity that you may or may not want to dive into. You will find a lot more biology from my cofounder. google scholar: chris anderson synthetic biology.
Question: could these be tailored for end-users? So instead a prescription a patient could be sent home with a jar of pickles that synthesized their medication as they fermented? I could see that disrupting the pharmaceutical industry (if dosage could be controlled).
This is really interesting - this field is new to me, but is there existing research available that shows that this works for large scale chemical production or will this be something that you'll be working on inventing?
Optimizing for large scale production is indeed what most of the field works on. There have been many success stories: an anti-malarial drug through yeast (Artemisinin by Amyris+Sanofi), a plastic precursor through bacteria (1,4-BDO by Genomatica+DuPont), and more.
What they are missing is the whole spectrum of what could be made biologically. We will create microbes for the most valuable chemicals and then partner with existing optimization companies that have industrial fermenters running. Think beer fermentation, just instead of the alcohol yeast, you use our yeast.
Really interesting, but the article doesn't explain why someone would want to grow chemicals using microbes instead of synthesizing them through chemistry. Is it cheaper?
Yes. When going after non-trivial chemicals the economics work out, and is cheaper when scaled up. The process essentially repurposes beer fermentation. The usual benefits of biological production: carbon-negative, and not using any toxic catalysts or producing waste products hold as well. Consider the molecule, paracetamol/acetaminophen/Tylenol [1], that we created a microbe for: The US alone produces 35000+ tonnes of it, all currently coming from petro-routes. It would be nicer to get it the same way we get beer, if we could.
There might be additional market forces that make a microbial fermentation attractive: e.g., in the case of the anti-malarial drug artemisinin, fluctuation in supply of the plant Artemisia annua, the price of the drug varied between $120-$1200/kg, and Amyris and Sanofi moved it to yeast based production for creating a steady supply [2]. That route was critical for the supply of the drug to African countries.
For many of the requests we get, chemical synthesis is not economically feasible to take the chemical to market, and bioproduction might be the only route. The fact that we suggest routes that work well for the planet is a good side-effect. :)
