OK, the “Silicon Valley Meets Biotech” subject has come up around here numerous times, most recently here, about a startup out of YCombinator called Verge Genomics. But several people have called my attention to this proposal over at (yes) YCombinator, so by gosh, it’s coming up again. Because this is just too much to believe.
It’s not strictly biopharma – this is a “request for a startup” to take a look at engineering plankton for better carbon sequestration. That’s not an insane idea per se, although it’s certainly a big risky one, but the worldview behind the proposal is perhaps the purest example of what I’ve called the “Silicon Valley Sunglasses” effect: the way that writing code and building hardware can give a person the illusion that everything else works the same way, that physical reality is there for you to re-engineer if you’re just smart enough and fast enough.
If we could figure out a way to solve phytoplankton’s mineral requirements, the entire ocean could become a powerful carbon sink, as well as a new frontier for economic growth. There are two ways of solving this problem – either get the missing minerals where they are needed (fertilization) or eliminate the need for minerals all together. We will start with what’s possible today and wade into what some might consider sci-fi. Either way, current biology won’t do the trick.
True. So far we are not in disagreement. The site starts off by talking about engineering phytoplankton to form “megastructures” to pull up more nutrients from the sea bed. How does one do that? Why, it’s just programming, you know:
When people talk about programming biology, they usually mean re-writing DNA to add some new feature. In this context, when we are talking about programming biology, we mean actually writing a complex, responsive biological programs with if/else statements, inputs and outputs that respond to a variety of different signals etc. We imagine something like a biological or biochemical Turing machine/computer instantiated at the level of a microbial cell. Actual programming. We imagine something like a more engineering or applications-oriented version of systems biology, where something approximating a multi-line python script could be instantiated in an organism.
That is actually what synthetic biology would like to achieve, and I agree that amazing things could be done when we get to that point. But no one is anywhere near it. Not within a million miles. We can hijack the natural systems inside cells, up to a point, and use the routines and structures that a billion or two years of evolution have left us. But there is no way to build a complex architecture from scratch. We don’t know how the ones we have work, and we don’t know how to make another that works, either. That paragraph is getting close to the Platonic ideal of the Valley view of how to do biology. I will leave for the reader to examine the attitude that we know enough about ecological systems to assume that turning plankton loose to make completely new megastructures in the ocean will work out fine because we clearly can work out the consequences beforehand.
But there’s more. You may be wondering about that earlier “eliminate the need for minerals” part, since those minerals are things like iron that are essential components of key enzymes. Or are they?
As we mentioned, mineral limitations are a major problem for using phytoplankton as a CDR tool. Many of these minerals are metals. These metals are used in various enzymes called metalloenzymes. All organisms have metalloenzymes. This may be leftover from life’s early evolution and may not necessarily be a necessity for biochemistry. For examples, the organisms Borrelia burgdorferi (lyme disease) and Lactobacillus plantarum (yogurt) have eliminated all iron from their metalloenzymes and instead use a different metal called manganese. This suggests that there is some flexibility with how metalloenzymes do their job. It might be possible to eliminate metals from these metalloenzymes all together.
Well. Some wheeled vehicles have spokes, while others have solid wheels, suggesting that it might be possible to eliminate wheels altogether and have cars that just sort of skid along the ground. You’ll get further in a wheel-less Tesla than you will with cells that have eliminated their requirements for all metal centers, I can tell you that. I don’t know quite where to start with this, only to suggest that this is the sort of idea that can only occur to a person who doesn’t actually know much chemistry. The reactivity of metal centers as they switch oxidation states would seem to be a very difficult thing to replicate by other means. I’m having trouble even imagining the beginnings of an idea, and honestly, I have a pretty vivid imagination.
Here’s the closest thing I can find to realism:
All of these approaches come with significant financial and technical risks. Even if all the technology was developed, it is entirely possible that the solution might not perform as expected. More modeling is required to get a higher resolution answer on this. In addition, all of these approaches rely on the release of genetically engineered phytoplankton into the ocean. Some folks might not be too keen on that idea.
That section does go on to mention that many of these things are currently beyond our abilities, but I detect a certain amount of “Well yeah, but the iPhone used to be beyond our abilities, too!” in there. What I really like is the idea that you can model your way to an answer for questions like these – that’s also an extremely revealing look into the worldview of the folks who wrote this. This is a sort of digital fundamentalism: everything comes back to 1s and 0s, because that’s all there is. Philosophically, would that be “Binism” instead of Monism? In the beginning was the bit, and it was capable of being a 1 or a 0, and Flipping of the Bit was the morning and evening of the first day? Hack the genome, hack biochemistry, hack ecology, hack the laws of chemistry and physics. We are not yet as gods, not even out in Mountain View.