Data Analysis

Q. How do I normalize my data?

A. Applied Microarrays strongly recommends whole array median normalization for all CodeLink whole genome bioarrays.  Descriptions of the relevant calculation methods are below, along with a description of other software calculations exported in the data sets.

Median Normalized Intensity: Intensity is defined as the sum of the pixel signals in the area of interest minus the scaled 2-pixel median background.  The intensities are then median-normalized by calculating the median of all spots (except for the control probes and blanks).  Each spot intensity is then divided by the median value.

Negative Control Threshold:  After median normalization, the negative control threshold is calculated using a set of predetermined negative controls.  Negative control threshold = (20% trim-mean of negative controls) + 3*standard deviation using the trim mean set.

\Coefficient of Variation (CV):  CV = (STDEV/MEAN). CV is generally displayed as a percentage.  For example, a (STDEV/MEAN) value of 0.08 corresponds to 8%.

Limit of Detection of Differential Expression:  This is the calculation of the minimal detectable spot ratio that can be distinguished from platform noise.  Only spots with intensities above the negative control threshold of the slide are used for minimum fold comparisons.  The percent within 2 fold lists the percentage of spots having ratios less than 2 for each slide-to-slide comparison.  The 95% within this fold result is the number at which 95% of the spot ratios fall below.  Thus, the lower the variability and noise of the system, the closer the ratios will get to the value of 1.

The Log2 Ratio calculation: Differential expression is reported as the log2 ratio of two spot values.  A spot with equal signals on each array would have a log2 ratio of zero, while a case of differential expression would result in a ratio value of 1 or greater.  A log ratio of 1 represents a 2-fold change.

Q. Why are there negative numbers for the intensity?

A. The "raw intensity" is calculated by taking the Mean Intensity of the pixels within the Spot minus the Median of the local background which is computed from the periphery of the spot.  If your spot has very little or no signal, the median of the local background may be slightly higher than the signal.  This condition will generate a negative number.   With regards to your analysis, if you encounter a negative number then you have little or no expression of that gene.

Q. How do I prepare my CodeLink™ data for GEO submissions?
A. Contact us at: This e-mail address is being protected from spambots. You need JavaScript enabled to view it and we will provide you with a CodeLink™ Expression Analysis Software patch to download.  Once installed, the patch will allow you to export in a tab-delimited text file for direct GEO submission.