BIOLOGICAL INSPIRATION FOR COMPUTING 297
This approach has two main drawbacks: the speed of individual assemblies, and the error rate.
First, the DNA reactions can take minutes or hours, and so any individual computation by self-assem-
bly will likely be substantially slower than using a traditional computer. The potential for self-assembly
is that, like exhaustive DNA computation, it can occur in parallel, with a parallelism factor as high as
1018. In the XOR experiment, researchers observed an error rate of 2 to 5 percent. Certainly, this rate may
be lowered as experience is gained in designing laboratory procedures and assembly methods; how-
ever, the error rate is likely to remain higher than that for electronic computers. For certain classes of
problems, an ultraparallel though unreliable approach may be an effective way to compute a solution.
8.4.3.3 Prospects
So far, DNA self-assembly has been demonstrated successfully in the laboratory, constructing rela-
tively simple patterns (e.g., alternating bands, or the encoding of a binary string) that are visible through
microscopy. It has also been used successfully for simple computations such as counting, XOR, and
addition.
Moving forward, laboratory techniques must improve in sophistication to handle the more complex
assemblies and reactions that will accompany large-scale computations or designs. Along with progress
in the lab, further theoretical developments are possible in developing algorithms for constructing
arbitrary aperiodic patterns.
Although so far DNA self-assembly has used only naturally occurring variants of DNA, a possible
improvement is to employ alternative chemistries, such as peptide nucleic acid, an artificial form of
DNA in which the backbone has peptide links in place of the phosphate that occurs in natural DNA.
Box 8.5
A Fantasy of Circuit FabricationConsider:... a fantasy of nanoscale circuit fabrication in a future technology. Imagine a family of primitive molecular-
electronic components, such as conductors, diodes, and switches, is available from generic parts suppliers.
Perhaps we have bottles of these common components in the freezer.... Suppose we have a circuit to imple-
ment. The first stage of the construction begins with the circuit and builds a layout incorporating the sizes of the
components and the ways they might interact. Next, the layout is analyzed to determine how to construct a
scaffold. Each branch is compiled into a collagen strut that links only to its selected targets. The struts are labeled
so that they bind only to the appropriate electrical component molecules. For each strut, the DNA sequence to
make that kind of strut is assembled, and a protocol is produced to insert the DNA into an appropriate cell. These
various custom parts are then synthesized by the transformed cells.
Finally, we create an appropriate mixture of these custom scaffold parts and generic electrical parts. Specially
programmed worker cells are added to the mixture to implement the circuit edifice we want. The worker cells
have complex programs, developed through amorphous computing technology. The programs control how the
workers perform their particular task of assembling the appropriate components in the appropriate patterns. With
a bit of sugar (to pay for their labor), the workers construct copies of our circuit we then collect, test, and package
for use.SOURCE: H. Abelson, R. Weiss, D. Allen, D. Coore, C. Hanson, G. Homsy, T.F. Knight, Jr., et al., “Amorphous Computing,” Commu-
nications of the ACM 43(5):74-82, 2000.