A New Procedure for Microarray Experiments to Account for Experimental Noise and the Uncertainty of Probe Response

Although microarrays are routine analysis tools in biomedical research, they still yield noisy output that often requires experimental confirmation. Many studies have aimed at optimizing probe design and statistical analysis to tackle this problem. However, less emphasis has been placed on controlling the noise inherent to the experimental approach. To address this problem, we investigate here a procedure that controls for such experimental variance and combine it with an assessment of probe performance. Two custom arrays were used to evaluate the procedure: one based on 25mer probes from an Affymetrix design and the other based on 60mer probes from an Agilent design. To assess experimental variance, all probes were replicated ten times. To assess probe performance, the probes were calibrated using a dilution series of target molecules and the signal response was fitted to an absorption model. We found that significant variance of the signal could be controlled by averaging across probes and removing probes that are nonresponsive. Thus, a more reliable signal could be obtained using our procedure than conventional approaches. We suggest that once an array is properly calibrated, absolute quantification of signals becomes straight forward, alleviating the need for normalization and reference hybridizations.