RelativeExpressionQuantificationAnalysis
Relative Expression Quantification Analysis
Overview
Computes relative expression using QBase method. The expression level histogram is viewable from the primer pair page. Normalization quality control results are viewable from the group main page.
Background
Expression level are computed for a primer pair across several samples from cycle threshold value (Ct) data read from the instrument. The level are relative to each other.
Relative expression levels (RE) are normalized with respect to a set of reference primer pairs (i.e. housekeeping genes), and allow inter-run calibration to correct for run-to-run variations from a selected calibrator sample repeated accross all runs. Reactions efficiencies are used for the cycle-to-quantity transformation (from primer pair efficiency analysis). Initial standard error on technical replicates and standard curve linear regression are propagated through all calculations and is refered to as the technical error. The technical error is different from the experimental error, which is the standard deviation of the RE values from a set of biological replicates. Technical replicates are automatically grouped by the sample naming scheme, whereas biological replicates needs a special annotation in the experiment data: n=i, where i is the replicate number.
Interpretation
RE histograms are viewable on the primer pair pages. Two different error bars are used to show both the technical and experimental errors: thin gray and bold black bars, respectively. Sample labels in histograms are suffixed by the number of replicate instances used to obtain the displayed value.
The normalization procedure is quality-controlled by quantifying the stability of the reference primer pairs used. The per-sample normalization factors (Nf) variation is measured by mapping tall pairwise ratio. These variations are a reflection of the different RNA input amounts in the experiments. The Nf ratio map is viewable at the group main page under NF pairwise rate map. If Nf variation are high (e.g. ratios above 6), then the expression bias due to the initial amounts of RNA used in the experiments is too strong to be removed. In such a case, it would be advisable to ignore the affected samples to increase the analysis quality on the remaining samples.
Stability of the reference primer pairs is also evaluated with respect to their expression level variations across all samples. Two statistics are used for this evaluation: the coefficent of variation (CV) and the sability parameters (M). CVs are calculated on the normalized RE of the reference primer pairs, hence CVs are expected to be low; typically below 0.250. Ms are calculated by averaging the pairwise RE variations, before normalization, to give an assessment on the independence of the reference primer pairs with respect to the samples used. A M value of 0 means perfectly equal RE across all samples; Ms are expected to be below 0.500. CVs and Ms are reported in a table at the group main page under Reports.
Algorithm
All relative expression calculation and quality control evaluation are implementations of the methods presented by Hellemans et al.:
- Hellemans J., Mortier G., De Paepe A., Speleman F., Vandesompele J. (2007) qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biol. 8:R19
Contacts
Credits
- Philippe Thibault
- Implementation of the analysis
