Biological chips go analog to boost efficiency
Analog chips run rings around digital ones at tiny scale
MIT boffins have figured out how to create synthetic analog organic circuits that can perform useful tasks without needing the sophistication that digital methods demand, which could lead to more efficient gloopy circuits and even more precise drug manufacture.
The advance, which could create technology for carefully managing protein creation for drug fabrication, was detailed in a paper entitled, "Synthetic Analog computation in living cells," which was published in the journal Nature on Wednesday.
"We have engineered synthetic circuits in living cells to perform complex computations such as add, subtract, divide, power laws, and logarithms. These computations are much more powerful and efficient than those previously described," researcher Timothy Lu told The Reg via email.
Analog DNA-based circuits have an advantage over digital DNA-based ones in that they don't require an abstraction that lets them deal in digital computation – for example, measuring the presence of a chemical and outputting 0 or 1.
Analog circuits don't require such binary translation, so they can perform some types of work better than their digital counterparts, and they can be energy thrifty as well.
"Cells are resource limited. A cell doesn't have the ability to have a billion parts in it," Lu told The Register. "If you've only got 4,000 parts [bits of DNA] to play with and all [are] performing simple zero to one computations, you can't really get a lot of complexity out of that. For analog you can perform computations like logs, ads, using much less parts than if you try and do it digitally."
To demonstrate their breakthrough, the researchers created circuits that summed protein inputs, calculated the log-transformed ratio of two different input inducers, and even worked out power laws.
The researchers' synthetic bio-chips were able to use in vivo (within-cell) binding functions to calculate power laws via analog computation. By comparison, "even one-bit full adders and subtractors in digital computation require several logic gates and, thus, numerous synthetic parts," the researchers wrote.
More efficient drug manufacturing
As for applications of the research, one would be to lower the cost of manufacturing drugs for pharmaceutical companies. "One of the things we're hoping to do is apply this in biotech," Lu said.
For instance, if a company wants to manufacture a protein within a cell, they need to walk a tightrope between creating (in biological terms "expressing") too much of the protein and damaging the cell, or being too cautious and making less protein than the cell can efficiently make. Finding this "happy medium" of maximum productivity is tricky, and is exactly what the researchers hope their analog circuits may be able to do.
"We're building a volume knob on your radio that can click through very many levels of expression quite easily and let you tune that in a particular way," Lu says.
Another way in which the research could be applied would be to create sophisticated analog control circuits as components of larger synthetic compute systems, they wrote. "More advanced systems may incorporate analog biosensors with feedback control of endogeneous genetic circuits to regulate phenotypes [in] a precise and dynamic fashion."
The research comes from a quartet of boffins at technology super-academy MIT. Previously some of the researchers had demonstrated that it is possible to bridge digital and analog computation with synthentic gene circuits in living cells.
"Thus, the natural extension of these precursors is to explore whether solely analog computation can also be performed," Timothy Lu told The Register. (Lu also led the team which announced in February that it had combined logic and memory within a single living cell for gloopy digital computation.)
Though the current analog research is "complementary" to the earlier work, "it doesn't build directly" on it, he said. "What we're suggesting here is, in addition to this type of computation, there's analog computation perhaps more suited to the physical requirements of cells." ®
Sponsored: Hyper-scale data management