Building computers out of the skeleton that holds our cells together instead of silicon could make them smaller and more energy efficient than the power-hungry machines we rely on today.
To encourage the development of such computers, Andrew Adamatsky at the University of the West of England in Bristol, UK, and his colleagues have produced a theoretical foundation for computing with cytoskeletons 鈥� the protein-based scaffolds that give cells their shape and help them move.
Cytoskeletons are made up of several different structures, including 25 nanometre-wide tubules made from a protein called tubulin and 6 nanometre-wide filaments made from a protein called actin. These structures regularly distribute information in the form of patterns of atoms, electrons and ions. By forcing these patterns to combine in various ways, it is possible to perform very basic computations 鈥� essentially producing the basic units of digital computers called logic gates.
Advertisement
Adamatsky has pulled all of this together into an overall theory for the first time. 鈥淚 fused the ideas in an attempt to develop a coherent theory of computing,鈥� he says.
He and his colleagues argue that cytoskeleton devices would have an advantage over ones made from DNA, another form of biological calculator that some researchers think could supercharge computing. DNA is good at storage but less good at processing signals, he says. Tubulin and actin are also less complex than DNA, which makes them easier materials to work with.
But Ross King at the University of Manchester, UK, who has worked on DNA computers, is sceptical. He thinks that it might be hard to get the various signalling mechanisms of cytoskeleton computers to produce deterministic decisions at the atomic level.
He agrees with Adamatsky that silicon鈥檚 days are numbered, however. We have reached the point where a single supercomputer consumes the same amount of energy as thousands of households, says King. 鈥淭hat鈥檚 not scalable.鈥�
Because cytoskeleton proteins and DNA molecules are so small they consume very little power. They can also be packed into a 3D structure more easily than silicon chips, which need to be laid out flat. This makes it easier to string millions of processors together so that they work in parallel.
Parallelism is powerful. Biological processors will never come close to matching the speed of today鈥檚 mainstream silicon chips. King says DNA can process about 100 instructions a second, for example 鈥� compared to many billions a second in your laptop. 鈥淏ut even though the actual processors are running quite slowly, you can get orders of magnitude more into a small space,鈥� he says. 鈥淵ou could have a desktop-sized supercomputer that used a fraction of the power.鈥�
Journal reference:听
