Quality measures for cDNA microarray experiments
by
Matthew Ritchie
The Walter and Eliza Hall Institute of Medical Research

As with any high throughput process, cDNA microarray experiments are sensitive to a multitude of production and experimental factors which can affect the quality of entire arrays or individual spots on arrays. It is desirable to 'flag' unusual spots and discard or down-weight them in further analysis to ensure robust results. Quality can be measured quantitatively at the spot level by empirical reproducibility. Typically, genes are printed in duplicate on arrays, and multiple arrays are involved in experiments, allowing reproducibility of the log(R/G) ratio to be explicitly measured. The extent to which spot specific measures such as area, intensity and signal-to-noise ratio are able to predict reproducibility in an experiment will be examined.

Date received: April 5, 2002