WHY THIS MATTERS IN BRIEF
The future of computing isn’t silicon, it isn’t even quantum, it’s biological, and it’s coming.
We’ve already turned bacteria into computing and storage devices that we’ve used to re-play videos from, and perform basic calculations, but just like tomorrow’s Biological computers DNA too, it’s alleged, is supposed to rescue us all from a computing rut. With advances using silicon petering out, DNA computers that can fit all of today’s computing power into a test tube, and that are more powerful than anything we have in the pipeline, including Quantum computers capable of running 100 million times faster than today’s computers, hold the promise of massive parallel computing architectures that are impossible today.
But there’s a problem – the molecular DNA “circuits” built so far, that we’re using to create the world’s first ultra-dense DNA storage devices that could store all the world’s information in a drive the size of a shoe box, all of which are being commercialised now, as well as the world’s first Artificial Intelligence neural network built out of DNA, have no flexibility at all.
Today, using DNA to compute is “like having to build a new computer out of new hardware just to run a new piece of software,” says computer scientist David Doty. So Doty, a professor at University of California Davis, and his colleagues set out to see what it would take to implement a DNA computer that was in fact reprogrammable. And as detailed in a paper published this week in Nature, Doty and his colleagues from Caltech and Maynooth University demonstrated just that.
They showed it’s possible to use a simple trigger to coax the same basic set of DNA molecules into implementing and running numerous different algorithms. Although this research is still exploratory, reprogrammable molecular algorithms like these could be used in the future to program DNA robots, like the ones created earlier this year, which have already successfully delivered drugs to cancerous cells.
“This is one of the landmark papers in the field,” says Thorsten-Lars Schmidt, an assistant professor for experimental biophysics at Kent State University who was not involved in the research. “[We had] algorithmic self-assembly before, but not to this degree of complexity.”
In electronic computers like the one you’re using to read this article, bits are the binary units of information that tell a computer what to do. They represent the discrete physical state of the underlying hardware, usually the presence or absence of an electrical current. These bits, or rather the electrical signals implementing them, are passed through circuits made up of logic gates, which perform an operation on one or more input bits and produce one bit as an output.
By combining these simple building blocks over and over, computers are able to run remarkably sophisticated programs. The idea behind DNA computing is to substitute electrical signals for chemical bonds and nucleic acids for silicon to create Bio-Molecular Software. According to Erik Winfree, a computer scientist at Caltech and a co-author of the paper, molecular algorithms leverage the natural information processing capacity baked into DNA, but rather than letting nature take the reins, he says, “computation controls the growth process.”
Over the past 20 years, several experiments have used molecular algorithms to do things like play tic-tac-toe or assemble various shapes. In each of these cases the DNA sequences had to be painstakingly designed to produce one specific algorithm that would generate the DNA structure. What’s different in this case is that the researchers designed a system where the same basic pieces of DNA can be ordered to arrange themselves to produce totally different algorithms – and therefore, totally different end products.
The process begins with DNA origami, a technique for folding a long piece of DNA into a desired shape. This folded piece of DNA serves as the “seed” that kickstarts the algorithmic assembly line, similar to the way a string dipped in sugar water acts as a seed when growing rock candy. The seed remains largely the same, regardless of the algorithm, with changes made to only a few small sequences within it for each new experiment.
Once the researchers have created the seed it is added to a solution of about 100 other DNA strands, known as DNA tiles. These tiles, each of which is composed of a unique arrangement of 42 nucleobases (the four basic biological compounds that make up DNA), are taken from a larger collection of 355 DNA tiles created by the researchers. To create a different algorithm, the researchers would choose a different set of starting tiles. So a molecular algorithm that implements a random walk requires a different group of DNA tiles than an algorithm used for counting. As these DNA tiles link up during the assembly process, they form a circuit that implements the chosen molecular algorithm on the input bits provided by the seed.
Using this system, the researchers created 21 different algorithms that could perform tasks like recognising multiples of three, electing a leader, generating patterns, and counting to 63. All of these algorithms were implemented using different combinations of the same 355 DNA tiles.
Writing code by dumping DNA tiles in a test tube is worlds away from the ease of typing on a keyboard, of course, but it represents a model for future iterations of flexible DNA computers. Indeed, if Doty, Winfree, and Woods have their way, the molecular programmers of tomorrow won’t even have to think about the underlying biomechanics of their programs, just like computer programmers today don’t need to understand the physics of transistors to write good software.
This experiment was basic science at its purest – a proof of concept that generated beautiful, albeit useless, results. But according to Petr Sulc, an assistant professor at Arizona State University’s Biodesign Institute who wasn’t involved in the research, the development of reprogrammable molecular algorithms for Nano-Manufacturing, and one day even molecular assemblers, the early prototypes of which are already emerging, the early opens the door for a wide range of potential applications.
Sulc suggested that this technique may one day be useful for the creation of nanoscale factories that assemble molecules or molecular robots for drug delivery. He said it may also contribute to the development of nanophotonic materials that could pave the way for computers based on light, rather than electrons.
“With these types of molecular algorithms, one day we might be able to assemble any complex object on a nanoscale level using a general programmable tile set, just as living cells can assemble into a bone cell or neuron cell just by selecting which proteins are expressed,” says Sulc.
The potential use cases of this nanoscale assembly technique boggle the mind, but these predictions are also based on our relatively limited understanding of the latent potential in the nanoscale world. After all, Alan Turing and the other progenitors of computer science could hardly have predicted the Internet, so perhaps some equally unfathomable applications for molecular computer science await us as well.