I’ve spent the last couple of years arguing that the barriers between software and the physical world are falling. The barriers between software and the living world are next.
At our Solid Conference last May, Carl Bass, Autodesk’s CEO, described the coming of generative design. Massive computing power, along with frictionless translation between digital and physical through devices like 3D scanners and CNC machines, will radically change the way we design the world around us. Instead of prototyping five versions of a chair through trial and error, you can use a computer to prototype and test a billion versions in a few hours, then fabricate it immediately. That scenario isn’t far off, Bass suggested, and it arises from a fluid relationship between real and virtual.
Biology is headed down the same path: with tools on both the input and output sides getting easier to use, materials getting easier to make, and plenty of computation in the middle, it’ll become the next way to translate between physical and digital. (Excitement has built to the degree that Solid co-chair Joi Ito suggested we change the name of our conference to “Solid and Squishy.”)
I spoke with Andrew Hessel, a distinguished research scientist in Autodesk’s Bio/Nano/Programmable Matter Group, about the promise of synthetic biology (and why Autodesk is interested in it). Hessel says the next generation of synthetic biology will be brought about by a blend of physical and virtual systems that make experimental iteration faster and processes more reliable.
On the input and output sides of experimentation — the physical portions — hardware is rapidly evolving. DNA sequencers have brought the cost of reading DNA into software down by more than three orders of magnitude since 2007 (vastly outpacing even Gordon Moore’s curve for computing power). The cost of writing DNA has also fallen dramatically, from about $4 per base pair in 2004 to between 5 and 10 cents per base pair today, says Hessel. Robotic laboratory systems from start-ups like Transcriptic and Emerald Cloud Lab make repetition and exploration easy, inexpensive, and reproducible.
In the middle, between DNA read-in and DNA read-out, is a computing layer that’s seen enormous growth in the availability of useful data. Databases like KEGG, for metabolic pathways, and Genbank, for genetic sequences, codify biological processes in machine terms. Software can instantaneously run through millions of permutations before suggesting a physical experiment. That’s the biological counterpart to Carl Bass’s algorithmic chair.
“What we’re seeing is the birth of a new programming industry.”
For decades, one of the exciting frontiers of biology has been the possibility of running experiments in silico as opposed to in vitro — simulating complex reactions and pathways with software and avoiding the expense and time needed to actually carry them out in a wet lab. Hessel says that movement is circling back toward physical implementation again: “In synthetic biology, it’s not just in silico anymore; you can print DNA so easily, put it into a living cell or cell system, run those experiments, and do living biology, too, alongside the computation,” he says. That sounds familiar to me from the new hardware movement, where electronics like the Arduino and Raspberry Pi have convinced a generation of programmers that they don’t need to stay locked behind their screens.
It’s easy to compare things to software, and the effect of doing so is always dramatic. Software is, after all, a pure, formal expression of human logic, and it has changed nearly everything in the modern world. I’m naturally a little skeptical of comparisons, then, but Hessel’s is very compelling:
“What we’re seeing is the birth of a new programming industry. The present digital computing industry started with the transistor, and you had to write the machine language at first — 0s and 1s. Now we have much more friendly programming languages, and computer programming is very sophisticated. Genetic engineering is similar, but instead of programming a computer, you’re programming a living cell, which is essentially a factory for biochemicals, and you have complete control over the production line. But today we’re still working at the machine-code level, and we’re just writing paragraphs. Eventually, as the technology gets better, we’ll be able to write whole books.”
Those “books” might include individually tailored pharmaceuticals, manufactured in distributed microfactories. If robotic biology labs and software can be shown to operate consistently and anticipate outcomes well enough, authorities could certify the processes to develop drugs, not the drugs themselves. Specialized microorganisms that help with environmental cleanup are another promising area, as are microorganisms that could produce fuel.
Ultimately, though, the scope of synthetic biology is the entire material world, and the field might soon join the toolbox of any creative person — one of several ways to translate between pure information and physical objects.
Cropped image on article and category pages by John Goode, on Flickr. Used under a Creative Commons license.
This post is part of a collaboration between O’Reilly and Autodesk. See our statement of editorial independence.