Research Article: Application of Equilibrium Models of Solution Hybridization to Microarray Design and Analysis

Date Published: June 10, 2010

Publisher: Public Library of Science

Author(s): Raad Z. Gharaibeh, Joshua M. Newton, Jennifer W. Weller, Cynthia J. Gibas, Arkady B. Khodursky.

Abstract: The probe percent bound value, calculated using multi-state equilibrium models of solution hybridization, is shown to be useful in understanding the hybridization behavior of microarray probes having 50 nucleotides, with and without mismatches. These longer oligonucleotides are in widespread use on microarrays, but there are few controlled studies of their interactions with mismatched targets compared to 25-mer based platforms.

50-mer oligonucleotides with centrally placed single, double and triple mismatches were spotted on an array. Over a range of target concentrations it was possible to discriminate binding to perfect matches and mismatches, and the type of mismatch could be predicted accurately in the concentration midrange (100 pM to 200 pM) using solution hybridization modeling methods. These results have implications for microarray design, optimization and analysis methods.

Our results highlight the importance of incorporating biophysical factors in both the design and the analysis of microarrays. Use of the probe “percent bound” value predicted by equilibrium models of hybridization is confirmed to be important for predicting and interpreting the behavior of long oligonucleotide arrays, as has been shown for short oligonucleotide arrays.

Partial Text: DNA microarrays [1] have revolutionized every area in biology [2]. Microarrays allow thousands of genes to be assayed at once, offering global views of biological processes at the transcriptional level [3], as well as allowing surveys of DNA sequence variation [2], and alternative splicing [4]. Integration of the results with other data informs many projects, such as those that perform cancer classification [5], genome annotation [6] and functional genomics [7]. The biology research community has invested heavily in microarray technology and values it, despite ongoing challenges with data quality and data interpretation.

In this study, we measured the response of targets binding to long oligonucleotide probes, designed in sets containing a small number of mismatches, over a range of target concentrations. Probes by their nature have different sequence composition, and consequently different thermodynamic properties, but affinity differences alone do not explain all of the observed variation. In particular, rules for predicting probe response at one concentration may not be valid at another concentration. One way to demonstrate this is to consider the results presented in Figures 1 and 2, in which probes containing mismatches show binding differences from the perfect match probes only at some target concentrations. The most effective model of the observed responses over a range of target concentrations incorporates a multistate equilibrium model of hybridization, the PPB. The use of a range of target concentrations to collect observations was very important in establishing the general applicability of this model: not only did it point to a weakness in the free energy model but it reflects the experimental reality that not all genes or genomic segments are present at the same concentration in most microarray experiments.