Nonequilibrium effects in DNA microarrays: a multiplatform study

TL;DR

This study investigates how DNA microarrays often do not reach thermal equilibrium within typical experimental times and proposes a method to detect this non-equilibrium state through fluorescence intensity analysis, supported by a kinetic model.

Contribution

It introduces a non-equilibrium detection method based on fluorescence intensity distribution and validates it across multiple microarray platforms using a 3-state hybridization model.

Findings

Deviations from equilibrium are observed in some microarray experiments.

The 3-state model predicts a proportionality of log intensity to free energy over an effective temperature.

Signal saturation occurs below equilibrium levels as predicted by the model.

Abstract

It has recently been shown that in some DNA microarrays the time needed to reach thermal equilibrium may largely exceed the typical experimental time, which is about 15h in standard protocols (Hooyberghs et al. Phys. Rev. E 81, 012901 (2010)). In this paper we discuss how this breakdown of thermodynamic equilibrium could be detected in microarray experiments without resorting to real time hybridization data, which are difficult to implement in standard experimental conditions. The method is based on the analysis of the distribution of fluorescence intensities I from different spots for probes carrying base mismatches. In thermal equilibrium and at sufficiently low concentrations, log I is expected to be linearly related to the hybridization free energy $\Delta G$ with a slope equal to $1/RT_{exp}$, where $T_{exp}$ is the experimental temperature and R is the gas constant. The breakdown of equilibrium results in the deviation from this law. A model for hybridization kinetics explaining the observed experimental behavior is discussed, the so-called 3-state model. It predicts that deviations from equilibrium yield a proportionality of $\log I$ to $\Delta G/RT_{eff}$. Here, $T_{eff}$ is an effective temperature, higher than the experimental one. This behavior is indeed observed in some experiments on Agilent arrays. We analyze experimental data from two other microarray platforms and discuss, on the basis of the results, the attainment of equilibrium in these cases. Interestingly, the same 3-state model predicts a (dynamical) saturation of the signal at values below the expected one at equilibrium.