Quantitative expression analysis using oligonucleotide microarrays based on a physico-chemical model

High-density DNA microarrays provide useful tools to analyze gene expression comprehensively. However, it is still difficult to obtain accurate expression levels from the observed microarray data because the signal intensity is affected by complicated factors involving probe---target hybridization, such as nonlinear behavior of hybridization, nonspecific hybridization, and folding of probe and target oligonucleotides. Various methods for microarray data analysis have been proposed to address this problem. In our previous report [7], we presented a benchmark analysis of probe---target hybridization using artificially synthesized oligonucleotides as targets, in which effect of nonspecific hybridization was negligible. The results showed that the preceding models explained the behavior of probe---target hybridization only within a narrow range of target concentrations. The experiments showed that finiteness of both probe and target molecules should be considered to understand detail behavior of hybridization.

In this paper, we present an extension of the Langmuir-model that reproduces the experimental results consistently and the 3-base nearest neighbor model to improve prediction accuracy. We also introduced effects of secondary structure formation, and dissociation of the probe---target duplex during washing after hybridization. The results will provide useful methods for the understanding and analysis of microarray experiments.