Reply to ‘Linking probe thermodynamics to microarray quantification’

We defend Langmuir-like models of microarrays from accusations by Li et al (2010 Phys. Biol. 7 048001) that they fail to link sequence-specific properties to hybridization signals. We argue that existing Langmuir-like models based on accepted principles of physical chemistry, together with a model of post-hybridization washing, are entirely consistent with various controlled experiments. Li et al’s competitive hybridization model on the other hand is not verified experimentally using designs which allow for an unambiguous differentiation with respect to Langmuir-like models and exhibits no benefit in fitting microarray probe intensities. Alternative models of microarray hybridization Plots of measured fluorescence intensities against spikein concentration for a given probe can be fitted well by a monotonically increasing, convex-downwards response function which asymptotes to a probe-sequence-dependent constant at high spike-in concentration. The majority of physical models proffered for explaining this response curve have their origins in Langmuir adsorption theory, which leads to a hyperbolic response function [10]. The central point of contention raised by Li et al [14] concerns the physical origin of differing saturation asymptotes of hybridization response curves in oligonucleotide microarray spike-in experiments. The Langmuir-based model cannot explain the probe-sequence dependence of this difference, but instead predicts that all probes must saturate completely to the same asymptote. Burden et al [8] and Held et al [11] have proposed that the problem is resolved by considering the posthybridization washing, which dissociates bound probes in a probe-sequence-specific manner. Subsequent experiments by Skvortsov et al [17] and Binder et al [3] are entirely consistent