Computing Comes to Life
Be Fruitful and Multiply
Persuading a living cell to perform useful computations is quite a trick, and yet it's not all that's needed. The real goal is to get a billion cells working in concert on the same task.
From one point of view, mass producing bacteria is extraordinarily easy. You don't have to build a billion-dollar "fab line" to manufacture them; just supply warmth and nutrients, and the bacteria will take care of proliferation on their own. The hard part is organizing a population of cells so that they work toward some specified goal. Here again electronic and biological technologies diverge. On a silicon chip, every circuit element has an assigned place and function, but living cells are squishy and motile and not easily confined to a rigid grid. The MIT group therefore takes another cue from biology, and lets large-scale structures emerge from processes akin to natural development and differentiation.
In an embryo, cells of identical genetic endowment differentiate into distinct tissues and organs, and also generate patterns such as the stripes and spots of animal pelts. What is most intriguing about biological development is that all the cells begin with the same "program," and they organize themselves without any externally designated leader. It all seems to be done by means of short-range communication between neighboring cells and the diffusion of chemical signals over longer distances. These same mechanisms would also be available to a multicellular biocomputer.
The study of large arrays of simple processing elements is a classical topic in computer science, but for the most part the arrays have been geometrically regular, and the processors have operated in strict synchrony. The MIT group offers a new paradigm of "amorphous computing" by spatially irregular and unsynchronized arrays. If all the processors run the same program, and they have only local communication, what patterns can emerge in such an amorphous blob of computers? Some of the examples generated so far have a distinctly botanical look to them, and yet they also resemble the design drawings for a silicon integrated circuit.
Many further hurdles remain before biocomputing could become a practical technology. Input and output are problematic. Maybe the input device will be a pipette of pheromone, and fluorescent proteins could produce output signals, but the expressive possibilities of these facilities seem rather limited. Large-capacity long-term storage—a biological disk drive—is also lacking. And speed is a concern, even with the extraordinary level of parallelism implicit in the exponential growth of a bacterial colony. Silicon processors are running at a gigahertz, but the speed of genetic circuits is in the millihertz range. Even a billion bacteria are no match for a Pentium.
But surely it would be a mistake to think of the E. coli computer as a beige box that will sit on your desk running a prokaryotic version of Microsoft Windows. A more likely prospect is a crop of programmable biological sensors, actuators and messengers. One contemplated application of such organisms is the assembly of nanoscale structures; instead of replacing semiconductor circuits, the cells would fabricate them. Another possibility is the old fantasy of a microscopic robot that could enter the human body to repair diseased tissues or combat infections. If this daydream of an intravenous computer is ever to happen, success seems more likely with the tools of genetic engineering than with a soldering iron.
Or maybe not. Maybe in the end it's just foolishness to imagine that anything so intellectually demanding as computation could be imposed on a biological substrate. No living organism can be expected to engage in abstract reasoning and symbol manipulation while carrying on with the daily routine of ingestion, growth, excretion, sleep, procreation. Get a life!
© Brian Hayes