Sensitivity and capacity of microarray encodings

Encoding and processing information in DNA-, RNA- and other biomolecule-based devices is an important topic in DNA-based computing with potentially important applications to fields such as bioinformatics, and, conceivably, microbiology and genetics. New methods to encode large data sets compactly on DNA chips has been recently proposed in (Garzon & Deaton, 2004) [18]. The method consists of shredding the data into short oligonucleotides and pouring it over a DNA chip with spots populated by copies of a basis set of noncrosshybridizing strands. In this paper, we provide an analysis of the sensitivity, robustness, and capacity of the encodings. First, we provide preliminary experimental evidence of the degree of variability of the representation and show that it can be made robust despite reaction conditions and the uncertainty of the hybridization chemistry in vitro. Based on these simulations, we provide an empirical estimate of the capacity of the representation to store information. Second, we present a new theoretical model to analyze and estimate the sensitivity and capacity of a given DNA chip for information discrimination. Finally, we briefly discuss some potential applications, such as genomic analysis, classification problems, and data mining of massive amounts of data in abiotic form without the onerous cost of massive synthesis of DNA strands. © Springer-Verlag Berlin Heidelberg 2006