Programming Wetware
Date: Aug 6, 2004
Press releases proclaiming "the world's smallest computer (about a trillion could fit in a drop of water)" were distributed this April by the public affairs department at the Weizmann Institute in Israel. Developed in the lab of Prof. Ehud Shapiro, this latest development in a series of biological computing innovations uses molecules within living cells to diagnose certain cancers and to produce drugs to combat them. The idea is to eventually release these tiny cellular machines inside the human body, where they would target cancer cells; the healthy cells would be left unmolested, assuming they are not destroyed by the body's immune system and they don't cause unintended side effects. This is progress, considering the original DNA test-tube computer created in Shapiro's lab in 2001 could do only simple calculations.
The idea of DNA based computers has been around for a decade or so. Leonard Adleman, a math professor at the University of Southern California (and the "A" in RSA encryption), published an article in Science magazine in November 1994, Molecular Computations of Solutions to Combinatorial Problems. He used a DNA based computer to solve the traveling salesman problem over seven cities. The virtue of a DNA computer, scientists say, is that you could have massively parallel problem solving (or encryption). "It launched the current government interest in biomolecular computing," commented Eric Eisenstadt, program manager at DARPA (the U.S. Defense Advanced Research Projects Agency).
The problem, Eisenstadt says, is that nobody knows how to create an algorithm in biological terms that effectively represents a real world optimization problem (such as a transportation problem). DARPA is thinking a lot about modeling and simulating organisms, particularly for chemical and biological detection. However, the agency is not getting involved in molecular computing, though "It is not for lack of trying," says Eisenstadt. "Engineering [organisms] for particular uses is quite plausible, plus it is practiced by big pharma. Computing still a toy idea."
However, the National Science Foundation actually has a big investment in this discipline, hoping to attack all sorts of difficult computational problems. NSF's Biological Information Technology and Systems (BITS) program will support research at the interface of biology and information technology.
Even Educated Fleas Do It
In another biotech discipline, dubbed "Synthetic Biology," scientists are reverse engineering cells from the ground up, constructing simple circuits from bio molecules, instead of deconstructing incredibly complex cells that communicate with one another in arcane and unknown ways. They hope to create simple organisms that can be programmed like machines by manipulating their DNA. Could these biological machines trump Moore's Law?
Let's take a look at what you'd need. In a biological computer, input, output, and "software" are composed of DNA. The "hardware" is formed by enzymes. By building a circuit out of lengths of DNA, cells can be induced to signal and communicate with each other. The goal is for a biological computer to do massively parallel calculations.
You also need standardized parts. MIT researchers are making Biobricks, circuits and components of DNA lengths that have specific functions and hook together within a cell like Lego blocks. These simple bacteriophages can be grown and stored separately, and interchanged as computing designs evolve.
One lab at MIT made clusters of E. coli cells that flash in unison. The researchers haven't managed yet to train the cells to spell out "Merry Christmas" under UV lights, but that is probably just a matter of time. Their next project is DNA Dots.
Should you don your latex gloves and learn to program wetware?
Biological computers (literally) have a few bugs that need to be worked out before they start appearing regularly in mail order catalogues. For one thing, while they can do rudimentary sort of calculation, input/output is exceedingly slow. That slimy blob would take a long time to do something as simple as balance your checkbook, and it would need regular feedings as well.
Silicon computers can switch between calculations. But you would have to construct a biological computer anew for each problem.
The electronics behind your computer chips run at almost the speed of light. Transistors are limited by "gating time," which is how long it takes the gate to open and close when you apply voltage. The gates of transistors composing chips now on the market are 130 nanometers (really small), which make them fast and power efficient.
But a biological computer is limited by diffusion, a relatively slow process. Plus, cells need a medium in which to grow. That biological computer could be a gooey mess.
Then there is the issue of competing microorganisms, or your computer getting moldy or viruses. A sterile environment is necessary, else you would have to worry that something won't eat your computer before your solution set is finished.
There would be other control issues. A biological device could, theoretically, stay alive indefinitely if it were fed constantly. Yet, every time a cell divides, there is an exchange of DNA. Errors creep in. Cells mutate. Your computer, after several generations, would not work. Plus, a biological device still would need an electronic read-write interface to interpret the data.
Nonetheless, techniques in synthetic biology promise useful applications in gene therapies. Examples include synthesizing drugs on a massive scale, such as artemisinin for malaria, and acting as sensors to detect toxins or explosives.
However, we still are a ways off from cellular slime doing long division in a dish.