Genetically engineered bacteria can work together to spot prime numbers, identify vowels or even work out the maximum number of slices a pizza can be cut into. The researchers behind the study say tiny biological computers like these can outcompete traditional computer chips in terms of both size and cost.
and his colleagues at the Saha Institute of Nuclear Physics in Kolkata, India, genetically engineered bacteria derived from Escherichia coli that could be combined in various ways to solve problems. These 鈥渂actoneurons鈥� were then arranged in various combinations in experiments to do 12 tasks.
鈥淚t鈥檚 a Lego-like system, it鈥檚 modular,鈥� says Bagh. 鈥淭hey鈥檙e not multicellular organisms, but they鈥檙e working together as a multicellular entity.鈥�
Advertisement
In a classical computer, zeros and ones 鈥� represented by low and high voltages 鈥� transmit information. But in bactoneurons, chemicals are used instead. Bagh says these engineered bacteria are reacting to a chemical stimulus, where binary numbers are input by the presence or absence of three chemicals representing ones or zeros.
Different arrangements of bactoneurons could show whether a digit between 0 and 9 was a prime number or if a letter between A and L was a vowel by fluorescing either green or red. They could even assess how many slices a pizza could be cut into using only straight lines.
The bactoneurons are each just 2 to 5 micrometres long, but they handle input and output and come with their own chemical power supply. Biological computers can be created in sizes far smaller than what is possible with classical computers, says Bagh. And because they are able to self-replicate, they could potentially be produced at scale and low cost.
Previous research on biological computers has created artificial neural networks from single-celled organisms to crunch a series of historical data points and forecast the future. But the researchers behind this new work say that until now there had been no demonstration of a programmable biological computer that can solve numerous problems.
The bacterial cultures work as a single-layer artificial neural network. This architecture is what lies at the heart of large language models like ChatGPT, albeit with far larger and more complex organisations of dozens or even hundreds of layers.
Nature Chemical Biology
