Bead numbers

The QuantiGene® Plex 2.0 platform utilizes the Luminex/xMAP magnetic bead array system to quantify multiple RNA targets simultaneously. Typically, QuantiGene output data is formatted as MFI values, median fluorescence across individual bead intensities. The individual bead intensities are usually not reported, which is why MFI values are considered raw QuantiGene data. The final number of magnetic capture beads per gene per well on readout is variable. Bead number values have been reported to range from 50-100 beads per gene per well (Ferrer, 2014) (Flagella, 2006), while the QuantiGene® 2.0 Plex assay user manual specifies an expected average bead count > 50 per gene (Affymetrix, manufacturer’s manual). Factors such as sample viscosity, washing steps throughout the assay and possible bead carryover across wells indeed affect the final number of beads per gene per well. Occasionally, this can drop to very low numbers in which case the corresponding MFI value cannot be considered a reliable median. Such values need to be discarded during the data analysis flow in order to ensure qualitative data. In a resampling experiment investigating 40 different genes from three different datasets we identified 40 as the minimum required number of beads per gene per well to guarantee stable and reliable MFI readout values (data currently not shown).
Raw MFI values obtained from a total number of beads < 40 are therefore not used while analysing expression levels. QGprofiler will list all gene-well combinations with a total number of beads < 40 and corresponding data values will be automatically discarded from downstream analysis.

Background controls 

Wells without loaded sample are referred to as background control wells. The average background signal is used for background correction. Additionally, gene-specific limits of detection (LOD) are calculated using the background signals. The LOD is defined as the average MFI value across background control wells plus three standard deviations of the background (Affymetrix, manufacturer’s manual). Any signals below this limit suggests absence of the corresponding gene and should not be used for the analysis of expression levels (should be set to zero?).
Three background control wells are suggested by the QuantiGene® 2.0 Plex user manual (Affymetrix, manufacturer’s manual). However, we recommend to increase this to at least six technical controls, based on experimental design (do we have data?). Moreover, QGprofiler will perform background correction and LOD calculation using the median background signal. This to ensure robustness of the calculated metric in case of outlying background signals. Additionally, plots on the background control wells are included in QGprofiler to facilitate visual quality control and to inspect variability and/or possible outlying values in the background control wells.

Assay linearity

Each QuantiGene® 2.0 Plex setup is defined by its probe set and biological sample (cellular context or tissue type). Every setup requires the identification of the optimal cell density or amount of tissue in order to ensure signals in the linear range of the assay. That is, a doubling of the cell density (or amount of tissue) should result in a doubling of the signal intensity. To this end, a serial dilution of sample should be run and observed signal ratios of background corrected MFI values should be calculated. Ratios within the 20% range of the expected ratio of 100% are accepted (QuantiGene® 2.0 Plex user manual). The final amount of cells (or tissue) should result in signals within the linear range for all genes in the probe set. Additionally, signals should be above the gene-specific limit of quantification (see 'Limit of Quantification'). In case the identified range is wide, expected effects on (the majority of) the genes could push the choice of optimal cell density (or amount of tissue) towards the lower or upper bound of the identified linear range.
QGprofiler includes plots on cell density to visually evaluate the linear range and aide in the decision process on the cell density/amount of sample in the real assay.

Limit of quantification

The limit of quantification (LOQ) refers to the lowest MFI value within the assay’s linear range (QuantiGene® 2.0 Plex user manual) and as such it is also gene-specific. The luminex/xMAP bead technology typically results in MFI values that level of at low/high treatment values, depending on the gene-specific effect. In order to ensure assay linearity in function of the chosen treatment, the LOQ is intended as a cut-off for low MFI values. It makes a call on the acceptability of MFI values to estimate true expression values. Only MFI values > LOQ are deemed to be useful in quantitative analyses of gene expression levels. However, given the throughput and the high number of measurable genes, determining the LOQ per gene, per experimental run is very cumbersome and in practice often not feasible. In this regard we propose a pragmatic cut off in analogy to the LOD calculation. This cut off can be set in a more relaxed or stringent manner, i.e. the median MFI value in background control wells plus five or ten standard deviations of the background. These are referred to as LOQ5 and LOQ10, respectively.
QGprofiler currently evaluates housekeeping genes against LOQ10 (see `Housekeeping genes`) and visually indicates observations of disease genes < LOQ10.

Housekeeping genes

Although QuantiGene® Plex 2.0 overcomes several of the pitfalls associated with classical mRNA quantification techniques such as qPCR, the need for internal normalization using housekeeping genes remains. It reduces variability in the results due to sample preparation, sample input and well/plate/experimental differences (QuantiGene® 2.0 Plex user manual). Any method relying on housekeeping genes for normalization should dedicate a vast portion to identifying truly stable housekeeping genes. The selection of such housekeeping genes in the context of the envisioned experiment is crucial to limit noise in the downstream data analysis and to ensure reliable results. In the case unstable housekeeping genes are used, the data transformation will result in unreliable relative expression and corresponding fold change values.
Proper housekeeping gene assessment is thus an important step in identifying the final probe set to be used in the QuantiGene setup. A suitable housekeeping gene exhibits constant expression levels across a wide range of doses and treatments in the cellular context (or tissue) that is envisioned in the assay setup. Once the final set of housekeeping genes for a particular assay is determined, it should remain constant throughout different experimental runs.
A housekeeping gene is said to be stable if its corresponding fold change values lie withing the [0.8; 1.2] interval, i.e. accepting a 20% variability. This is in analogy with the rationale to accept a 20% deviation from the expected 100% ratio when determining the linear range of the assay. A housekeeping gene should be stable across different experimental conditions (treatments and doses) in order to serve as normalization gene. We recommend the inclusion of at least three different stable housekeeping genes in the final probe set..
QGprofiler includes plot on FC values of housekeeping genes to evaluate stability using the 100% +/- 20% ruleof the assay linearity logic...