QC Interpretations of the Metrics.txt file

Under the "Image Data" folder, along with the TIF, IDAT and LOCS files, there is a "Metrics.txt" file that summarize a few QC indicators during the image scanning process.

Of interest is the "RegGrn", which runs from 0 (bad) to 1 (good), to indicate if the image has been registered properly. Note that the "RegRed" is irrelevant for expression arrays, because the expression arrays are single color.

Although a value of 0 on the "RegGrn" should suggest a problem, Dr. Wei Shi (WEHI) in a recent Bioconductor post suggested that sometimes the data looks fine when RegGrn is close to zero. Dr. Shi quoted an answer from Illumina technical support on this issue:

The 0 in the metrics file may indicate that there is a problem with
registration of the stripe but there are other ways to look at this.
First, have you looked at the data from the Beadchip in BeadStudio?
Does the data appear consistent with the rest of the samples in your
experiment? In addition, do the controls look OK for this sample when
compared to the other samples?

Next, you can look at the registration visually in BeadStudio as long as
you have saved tif images when you scanned the BeadChip. To do so, go
to the Analysis menu in BeadStudio and select View Image then choose the
stripe you are interested in. Now select Overlay Cores (the icon looks
like 3 blue-green circles) and look at the image to see if the green
circles line up with the intensity spots on the image. If the BeadChip
is not registered the circles representing cores will not line up with
the spots on the array and the circles themselves will be somewhat
disordered.

If you have reason to suspect that registration is affecting your data,
you can rescan the BeadChip. Gene expression BeadChips can be rescanned
as long as you do so within a few days and if the BeadChips have been
stored in the dark. If you do recan the BeadChip, please uncheck "Save
Compressed Images"; this will allow you to save tif images.