08.01.2013 Views

LNCS 2950 - Aspects of Molecular Computing (Frontmatter Pages)

LNCS 2950 - Aspects of Molecular Computing (Frontmatter Pages)

LNCS 2950 - Aspects of Molecular Computing (Frontmatter Pages)

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Digital Information Encoding on DNA 159<br />

can be done so that the composition <strong>of</strong> the strands follows a predetermined<br />

pattern (e.g., 60% c/g’s and 40% a/t’s). All biners are then uniformly mapped to<br />

aDNAcodewordbythesameassignment.Theresultingcodethushasminimum<br />

h-distance between DNA codewords no smaller than in the original BNA seed.<br />

Fig. 2(b) shows the quality <strong>of</strong> the codes obtained by this method, to be discussed<br />

next.<br />

3.4 Evaluation <strong>of</strong> Code Quality<br />

For biner codes, a first criterion is the distribution <strong>of</strong> h-distance between codewords.<br />

Fig. 1(a) shows this distribution for the various codes generated by tensor<br />

products. The average minimun pairwise distance is, on the average, about 33%<br />

<strong>of</strong> the strand length. This is comparable with the minimum distance <strong>of</strong> l/3 in<br />

template codes.<br />

As mentioned in [2], however, the primary criterion <strong>of</strong> quality for codeword<br />

sets is performance in the test tube. A reliable predictor <strong>of</strong> such performance is<br />

the Gibbs energy <strong>of</strong> hybridization. A computational method to compute the<br />

Gibbs energy valid for short oligonucleotides (up to 60 bps) is given in [8].<br />

Code sets obtained by generate-and-filter methods using this model have been<br />

tested in the tube with good experimental results [8,10]. The same Gibbs energy<br />

calculation was used to evaluate and compare the quality <strong>of</strong> code sets generated<br />

by the template and tensor product methods. According to this method, the<br />

highest Gibbs energy allowable between two codewords must be, at the very<br />

least, −6 Kcal/mole if they are to remain noncrosshybridizing [8].<br />

Statistics on the Gibbs energy <strong>of</strong> hybridization are shown in Fig. 2. The template<br />

codes <strong>of</strong> very short lengths in Fig. 2(a) remain within a safe region (above<br />

−6 kCal/Mole), but they have too low Gibbs energies that permit crosshybridization<br />

for longer lengths. On the other hand, although the tensor product operation<br />

initially decreases the Gibbs energy, the Gibbs energies eventually stabilize and<br />

remain bounded within a safe region that does not permit cross hybridization,<br />

both on the average and minimum energies.<br />

Fig. 3 shows a more detailed comparison <strong>of</strong> the frequency <strong>of</strong> codewords in<br />

each <strong>of</strong> the two types <strong>of</strong> codes with a Gibbs energy prone to cross hybridization<br />

(third bar from top). Over 80% <strong>of</strong> the pairwise energies are noncrosshybridizing<br />

for template codes (top and middle bars), but they dip below the limit in<br />

a significant proportion. On the other hand, all pairwise energies predict no<br />

crosshybridization among tensor product codewords, which also exhibit more<br />

positive energies. Finally, the number <strong>of</strong> codewords produced is larger with tensor<br />

products, and increases rapidly when the lengths <strong>of</strong> the codewords reaches<br />

64− and 128−mers (for which no template codes are available.)<br />

4 Encoding Abiotic Information in DNA Spaces<br />

This section explores the broader problem <strong>of</strong> encoding abiotic information in<br />

DNA for storage and processing. The methods presented in the previous section

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!