Figure 2

Unreliable Spot Finding. Scatter plot of intensities from slides of two print batches with the same high antibody concentration on a binary logarithm scale. Due to experimental noise, we see departures from the diagonal line on which we would expect all data points. The quality control algorithm identifies approximately 2.5% of all data points in each print batch as unreliable across all subarrays; these peptides are removed accordingly (colored in magenta, cyan and orange), resulting in an increase of the coefficient of variation of approximately 3%. While not identifying all outlying observation, the removed spots primarly affect peptide spots which show large variation between the print batches.