The within-array quality for (genomic) microarrays is often measured using the following metrics:
- Standard Deviation Autosome / Robust (SD autosome): Measure of the dispersion of Log2 ratio of all clones on the array, giving an overall picture of the noise in the array. It is calculated on the normalised but un-smoothed data. The SD robust is the middle 58%/66% of the data. By excluding outliers large changes such as trisomies will not cause this number to change significantly. (The SD robust is the number we use when we say “3 SDs away from the noise” in the calling algorithm.) Both measures are given after all data processing but excluding any smoothing. For BlueFuse Multi processed data the values should be 0.07-0.15 and 0.05-0.11 for the autosome and robust measure respectively.
- Signal to Background Ratio (SBR): Brightness of the mean signal (after the background has been subtracted) divided by the raw background signal (global signal).
- Derivative Log2 Ratio / Fused (DLR): Measure of the probe-to-probe variability. In an ideal world, probes within a region will have essentially the same ratio. In a noisy array adjacent probes can have a very large ratio difference. The DLR raw is before any data processing, DLR fused is after normalization and data correction BUT is always done on un-smoothed data so it is user setting independent and a cannot be adjusted by the user thereby giving a consistent array-to-array measure of noise. BlueFuse results should be < 0.2.
- % included clones: Percentage of all clones that were not excluded on a BAC array due to inconsistencies between clone replicates. For BlueFuse results this should be > 95 %.
- Mean Spot Amplitude: The mean fluorescent signal intensities for the two channels; channel 1 = sample (standardly Cy3; ex 550nm, emm 570nm) and channel 2 = reference (standardly Cy5; ex 650nm, emm 670nm). This metric is variable due to the differences between available scanners. The mean spot amplitude metric can give an indication of how well the DNA has labelled with fluorescent dyes, but more importantly, really high values can indicate over scanning of the microarray image OR can indicate poor washing so there is lots of non-specific signal left. The balance between channels can be assessed but the Cy5 signal tends to give a higher intensity than Cy3, major differences in the channels may indicate a labelling or a scanner problem.
Source: BlueGnome user docs
See also: Microarray Scanners and PGS consulting in the UK & Ireland