Theoretical and Experimental DNA Computation (Natural ...

5.8 Experimental Investigations 133
Experiment 21. Evidence of exclusion was seen again, but in all cases it was
incomplete.
Experiment 22. There was evidence of specific exclusion (the intensity of
targeted sequences reduced) but the process was incomplete. There seemed
to be a basic problem with the method in that it used the enzymatic removal
process to target and destroy specific sequences, followed by an incredibly
sensitive technique to detect them.
Experiment 32. The PCR failed to produce any product from any of the
samples, including the positive control. This was probably due to loss of the
template during the washing steps, reducing its concentration below the limit
of detection.
Discussion
It is perhaps useful at this point to note that this implementation, first proposed
in [12], established for the first time the “destructive” DNA-based algorithmic
paradigm, which has been subsequently used in several groundbreaking
papers [58, 99].
In [141] Seeman et al. describe the potential pitfalls that may confront
experimentalists working on DNA computation. In this section we describe
in a similar fashion the lessons to be drawn from the experimental investigations
just described. We hope that other experimentalists in the field may be
made aware of various subtle aspects of the implementation of models of DNA
computation. We have found that the requirements of DNA-based algorithmic
experiments are often more strict than those of “traditional” investigations
in molecular biology. For example, it is rare that molecular biologists are
required to sequence a heterogeneous population of DNA strands; yet, for
any nontrivial problem, this task is inevitably required as the final step of
the implementation of a DNA-based algorithm. We hope that these (often
non-obvious) impediments to efficient and error-resistant implementation of
models of DNA computation will be made apparent in the following sections.
Ensure appropriate control and optimization experiments are
performed
We quickly found that a major component of the work was comprised of
finding optimal experimental conditions. Factors to be taken into account
included strand concentration, salt concentration, restriction enzyme concentration,
annealing temperature, and number of cycles. Due to the unusual
nature of the experiments, we found that the system was far more sensitive
to experimental conditions than is normally the case.