# Comparison of normalization methods for Illumina BeadChip HumanHT-12 v3

- Ramona Schmid†
^{1}, - Patrick Baum†
^{1}, - Carina Ittrich
^{1}, - Katrin Fundel-Clemens
^{1}, - Wolfgang Huber
^{2}, - Benedikt Brors
^{3}, - Roland Eils
^{3, 4}, - Andreas Weith
^{1}, - Detlev Mennerich
^{1, 5}and - Karsten Quast
^{1}Email author

**11**:349

**DOI: **10.1186/1471-2164-11-349

© Schmid et al; licensee BioMed Central Ltd. 2010

**Received: **9 March 2010

**Accepted: **2 June 2010

**Published: **2 June 2010

## Abstract

### Background

Normalization of microarrays is a standard practice to account for and minimize effects which are not due to the controlled factors in an experiment. There is an overwhelming number of different methods that can be applied, none of which is ideally suited for all experimental designs. Thus, it is important to identify a normalization method appropriate for the experimental setup under consideration that is neither too negligent nor too stringent. Major aim is to derive optimal results from the underlying experiment. Comparisons of different normalization methods have already been conducted, none of which, to our knowledge, comparing more than a handful of methods.

### Results

In the present study, 25 different ways of pre-processing Illumina Sentrix BeadChip array data are compared. Among others, methods provided by the BeadStudio software are taken into account. Looking at different statistical measures, we point out the ideal versus the actual observations. Additionally, we compare qRT-PCR measurements of transcripts from different ranges of expression intensities to the respective normalized values of the microarray data. Taking together all different kinds of measures, the ideal method for our dataset is identified.

### Conclusions

Pre-processing of microarray gene expression experiments has been shown to influence further downstream analysis to a great extent and thus has to be carefully chosen based on the design of the experiment. This study provides a recommendation for deciding which normalization method is best suited for a particular experimental setup.

## Background

Analysing gene expression using microarrays is a well established method [1]. Many different technologies have been developed, of which the most advanced are Affymetrix GeneChip [2] and Illumina Sentrix BeadChip arrays [3]. These high throughput technologies allow the parallel quantification of a large number of transcripts. It is well known in the microarray community that normalization has to be performed to minimize systematic effects that are not constant between different samples of an experiment and that are not due to the factors under investigation (e.g. treatment, time).

Several studies comparing different normalization methods have already been conducted, many of them focusing on Affymetrix chips [4–7], others on Illumina chips [8–12], and only very few have been conducted focusing on both technologies [13, 14]. To our knowledge, so far no analysis has been published comparing a large number of different normalization methods for the Illumina BeadChip Technology and only very few studies [8] that took the normalizations offered by BeadStudio into account and compared them to other established normalization methods. Optimal selection of a normalization method depends very heavily on the nature of the experiment. In this regard factors like comparability and quality of single runs play a major role. It has been shown that the normalization method used may influence further downstream analysis to a great extent [6] and thus has to be carefully chosen based on the actual data.

Here we present a strategy for an in depth evaluation of normalization methods aiming at identifying the most appropriate one for a given data set. Our study compares established normalization methods available in the R environment to those offered by BeadStudio software. It focuses on the HumanHT-12 v3 Expression BeadChip, yet the underlying principles are directly transferable to other technologies measuring gene expression. Analyses described here provide the basis for the Phenocopy project (Baum *et al*., Phenocopy - a strategy to qualify chemical compounds during Hit-to-Lead and/or Lead Optimization, submitted 2010). Within this project, the aim is to predict mode of action as well as off-target effects of compounds based on gene expression data. To do so, HaCaT cells were treated with TGF-β as well as seven different compounds in seven dosages and with six siRNAs. Gene expression was measured after 2, 4, and 12 hours using Illumina's Human HT-12 v3 Expression BeadChips [15]. In addition we performed qRT-PCRs using the TaqMan^{®} technology to measure the quantitative abundance of eight genes at three time points that are known to be deregulated [11, 14]. In order to identify the normalization method best suited for the experimental design of the Phenocopy project, we, in total, compared 25 different ways of normalization and analysed different statistical aspects of the data.

## Results and Discussion

_{2}-transformation (log) or variance-stabilizing transformation (vst) [9]. Since BeadStudio's background normalization can lead to negative values, the data had to be transformed to contain only positive values by using either the background correction of rma[4] or forcePos[17] to be able to apply log

_{2}-transformation. In a last step, the data was normalized using quantile, loess, or rsn[17] normalization. Alternatively, the transformation steps were skipped and vsn[18] or the normalization methods supplied by BeadStudio (average, rankInvariant, cubicSpline) are used for normalization.

Different pre-processing methods were evaluated by analysing the variance of the resulting gene expression intensities via various statistical measures. Some of these have already been used in other studies [5]. In addition to the investigation of the actual expression intensities, fold changes derived from resulting gene expression intensities were compared to fold changes based on quantitative measurements of RNA abundance as determined by qRT-PCR. Thereby, it is possible to evaluate the pre-processing methods with respect to their bias.

Pre-processing methods were scored from -2 to 2 based on how well they match the required criteria for the different analyses described in this section. As it is difficult to clearly categorize the methods based on the examined measures, the final decision of which score to assign to some extent stays subjective. However, it is unambiguously possible to separate better pre-processing methods from worse. A complete overview of the scores assigned and the final ranking is given in Figure 1.

### Analyses of variance based on expression measurements

One basic assumption of gene expression pre-processing methods is that the majority of genes do not change their expression under different conditions. Additionally, expression intensities of replicates should be very similar compared to the expression of transcripts between differently treated sample groups. Based on these principles, we looked at different statistical measures to identify the method best suited for our dataset with respect to variance.

#### Distribution of F-test statistics

#### P-values against variance between groups

_{10}(p-value) should be. When plotting these parameters, an appropriate normalization method should result in smoothly increasing values with not too much scattering around the fitted curve. Figure 3 displays the -log

_{10}(p-value) against the respective variance between the control groups at time points 2 h, 4 h, and 12 h for three of the pre-processing methods, an overview over all results is given in Additional file 1. Normalizations reflecting the described properties are for example noBg_cubicSpline, noBg_log_rsn, noBg_vst_rsn, and noBg_vsn. All of the normalizations performed on rma background corrected data as well as bg_vsn display a relatively high -log

_{10}(p-value) for a relatively high proportion of low between group variability values leading to a high scattering of observations in these regions. Using, for example, the rank invariant normalization of BeadStudio (noBg_rankInvariant) the p-values for the low between group variability tend to be relatively small. This could lead to an overestimation of differentially expressed genes when filtering solely based on p-values.

#### Boxplots of MSQ_{between} and MSQ_{within}

_{between}) and within (MSQ

_{within}) group variances and their relation to each other. If genes are not differentially expressed, MSQ

_{between}should be comparable to MSQ

_{within}. For genes that are differentially expressed, MSQ

_{between}is supposed to be higher than MSQ

_{within}. Figure 4 displays the boxplots for MSQ

_{between}(red) and MSQ

_{within}(blue) values. Since we expect some genes to be differentially deregulated across the different time points under consideration, quantiles of MSQ

_{within}values should lie below the corresponding quantiles of the MSQ

_{between}values. For the differentially expressed genes, within group variance should be smaller than between group variance, whereas for the genes not differentially expressed, the respective MSQ

_{between}and MSQ

_{within}values should show no great difference. Small interquartile ranges (IQRs) of MSQ

_{within}are indicative for a comparable variability between genes.

To judge the values for MSQ, an MSQ_{between} was calculated for artificial group means of log_{2} expression values for three time points based on four replicates. The group means used were (6, 6, 7) which resulted in an MSQ_{between} of 1.33, indicated by a dashed grey line in Figure 4. The mean expression values of the artificial groups have been chosen such that they exhibit a log_{2} ratio of 1 when group 3 is compared to group 1 or group 2, reflecting a relevant difference between those groups. A good normalization method should result in similar expression values for replicates and thus in small MSQ_{within} values hardly crossing this artificial MSQ_{between}. Additionally, since we limited the whole data set to expressions measured for untreated HaCaT cells across time, we expect only few genes to be differentially expressed. Thus, only a few genes are assumed to result in an MSQ_{between} above the artificial MSQ_{between}.

Almost all boxplots representing MSQ_{within} of background normalized data (bg_*) result in outliers crossing the artificial MSQ_{between}, only those transformed using vst do stay below. Compared to other pre-processing methods, noBg_vst_loess, noBg_log_loess, and bg_vst_loess show a relatively wide IQR for both, MSQ_{between} and MSQ_{within}. Methods that meet the described behaviour by showing a low within group variability for which the quantiles generally exhibit lower values than the quantiles of the between group variabilities are bg_vst_quantile, bg_vst_rsn, noBg_log_quantile, noBg_log_rsn, noBg_noNorm, noBg_vsn, noBg_vst_quantile, and noBg_vst_rsn.

#### Density functions of MSQ_{between} and MSQ_{within}

Density functions of MSQ_{between} and MSQ_{within} should exhibit clear differences. This fact renders density functions of MSQ_{between} and MSQ_{within} as an additional option for investigating these values. Within group variability should be smaller than between group variability and most of the genes should show a between group variability similar to the within group variability, i.e. are not differentially expressed. Thus, the mode of MSQ_{within} should be smaller than the mode of MSQ_{between} and the peak of the function for MSQ_{within} is supposed to be higher than the peak for MSQ_{between}. Lean MSQ_{within} functions, on the one hand, reflect a comparable within group variability for many genes. On the other hand, broader MSQ_{between} functions indicate that at least some of the genes, i.e. the differentially expressed ones, show a higher between than within group variability. Ideal characteristics of density functions as described here are very similar to the characteristics of ideal boxplots mentioned in the previous section. In contrast to density functions, boxplots give a very rough idea about the distribution of the values, also depicting outliers. Density functions deliver a more detailed view of how the values are distributed across different ranges.

_{between}(red) and MSQ

_{within}(blue) for three of the pre-processing methods, a complete overview is given in Additional file 2. In particular density plots representing the normalization methods noBg_log_quantile, noBg_log_rsn, and noBg_vsn best exhibit the desired behaviour. Unexpectedly the density functions of MSQ

_{within}generated by bg_vst_loess, noBG_log_loess, and noBg_vst_loess are bimodal. One reason for bimodal density functions could be a group of transcripts exhibiting higher variability compared to other transcripts. In general, it is expected that the data shows a consistent variability. Having the opportunity to choose between normalization methods resulting in unimodal or bimodal density functions for MSQ

_{within}, normalization methods leading to a unimodal distribution should be favoured.

A small overlap of the functions like for the values generated by the noBg_average normalization (Additional file 2) would indicate the unlikely event that most of the genes show a higher between than within group variability, i.e. are differentially expressed. Thus, this normalization method is probably not adequate.

#### Volcano plots

_{10}(p-value) versus the respective log

_{2}ratios. Due to the tendency of larger log

_{2}ratios being connected to more significant -log

_{10}(p-values) a volcano like shape is generated. Pairwise comparisons (4 hours compared to 2 hours, 12 hours compared to 2 hours, and 12 hours compared to 4 hours) using a moderated t-statistic [21] were performed to calculate log

_{2}ratios and p-values. Our aim is to detect normalization procedures yielding as correct estimates of log

_{2}ratios as possible combined with as informative p-values as possible. As mentioned above higher log

_{2}ratios should tend to have higher -log

_{10}(p-value). The loess fits of the log

_{2}ratios and -log

_{10}(p-value) pairs (dark blue curves) of the volcano plots shown in Figure 6 and Additional file 3 shall neither be too flat nor too narrow and the scatter of the p-values for specific log

_{2}ratios should not be too large.

All volcano plots based on rma background corrected data do not look very promising. The fitted curves are rather flat, i.e. even for high absolute log_{2} ratios the -log_{10}(p-value) are relatively low. Additionally, the -log_{10}(p-value) for similar log_{2} ratios tend to scatter extremely. Some volcanos, e.g. bg_average, bg_noNorm, bg_rankInvariant, and noBg_rankInvariant, show an unsymmetrical relation between p-values for negative and positive log_{2} ratios. Especially noBg_rankInvariant exhibits a bias towards small negative log_{2} ratios for which the respective -log_{10}(p-value) seem to be relatively high. In this region the fitted curve shows a very steep, linear course. Volcano plots generated for all other methods are similar to what would be expected. Still they differ in the variance of the p-values and in that some of the fitted curves show a flatter shape than others. This reflects the fact that some normalization methods generate a smaller variance than others, resulting in lower fold changes but more significant p-values. Ultimately, a method with a reasonable trade off between fold change and variance has to be chosen and cut-off parameters for interesting genes have to be defined accordingly. Volcano plots best reflecting the desired properties in the context of our experiment were generated by noBg_log_quantile, noBg_log_rsn, and noBg_vsn. They show the least scattering of values around the fitted curves, but they probably underestimate fold changes. Plots produced bye bg_forcePos_log_loss, bg_forcePos_log_quantile, bg_forcePos_log_rsn, bg_vst_quantile, bg_vst_rsn, noBg_cubicSpline, noBg_log_loess, noBg_vst_loess, noBg_vst_quantile, and noBg_vst_rsn also fulfil the above mentioned criteria, but show more scattering.

#### Residual standard deviation against mean and minimum of gene expression levels

In an optimally normalized experiment the residual standard deviation of fitted gene expression intensities should be low and independent of the expression levels, i.e. the variance over the different expression levels should be stable. This is prerequisite for many statistical methods, like for example linear model fitting and moderate t-statistics [21], that are utilised for analysing gene expression data.

Methods which perform best with respect to variance stabilization across all expression levels are bg_vst_loess, bg_vst_quantile, bg_vst_rsn, noBg_vst_loess, noBg_vst_quantile, and noBg_vst_rsn.

#### Scatterplots of expression values

#### Pseudo-ROC curves

### Analyses of bias based on qRT-PCR

qRT-PCR has been performed for mRNAs from eight genes that are known to be deregulated by TGF-β signalling to a varying degree (CDKN1A, CDKN2B, HAND1, JUNB, LINCR, RPTN, SERPINE1, and TSC22D1). By this means, it is possible to compare the results of the normalization methods to values that reflect the real abundance of the respective mRNA in the cells. Thus, we are able to evaluate the accuracy of the different pre-processing methods with respect to their bias. To guarantee that the comparisons of the normalization methods are not biased towards certain intensities, the mRNAs used in qRT-PCR experiments were chosen such that the respective signals on the chips cover a broad range of expression intensities (Additional file 8).

#### Correlation analysis of fold changes

_{2}-transformation (bg_rma_log_*), cubic spline normalization (bg_cubicSpline), or variance stabilizing normalization (bg_vsn).

#### Regression analysis

_{2}-transformation (bg_rma_log_*) best fit the qRT-PCR results (Figure 12).

## Conclusions

It is important to select appropriate pre-processing methods for a given data set based on the experimental setup used. On the one hand, if sample sizes of the different groups are relatively small, it is crucial to achieve a homogeneous variance for the groups. On the other hand, if sample sizes are large, variances can be estimated separately and one should focus on unbiased fold changes. Since the sample sizes for the current data set are rather small (three to four replicates per group), a stable variance is more important than an exact representation of the fold change. In general, the data should be normalized without too much reducing real variations. Figure 1 summarizes the quality measures for all methods we investigated, demonstrating the background for the final choice. Clustering of the quality scores assigned reveals two major tendencies based on background normalization. On the one hand, data that was background normalized (bg_*) tend to better reflect the real fold changes, i.e. show less bias. On the other hand, pre-processing without background normalization (noBg_*) leads to a more homogeneous variance. Accurately defined, constant experimental conditions across all experiments as well as their parallel conduction probably have lead to a relatively consistent background level across all samples. Since background correction could introduce additional variation, these could be the reasons why, for our data set, data that was not background normalized (noBg_*) in general provides better stabilization of variance than background normalized data (bg_*). Methods combining background normalization with vst (bg_vst_*) constitute an exception. Here, vst leads to a better stabilization of variance while introducing more bias. As vst estimates an offset for the background based on the data [9], noBg_vst_* and bg_vst_* pre-processing methods could lead to similar results.

One has to keep in mind that, based on the individual analyses, there are several methods resulting in nearly equal quality. Therefore, it is not possible to give a well-defined rationale for using only one specific method. After excluding the methods that clearly violate the imposed criteria, the decision is still subjective. It, for example, depends on whether one would like to account for a good estimate of fold changes or a small and homogeneous variance. Finally the decision remains based on experience; yet, with the analyses and criteria described here, we provide a recommendation on how to pre-select appropriate methods. Since, for our data set, we intended to achieve a low and homogeneous variance, we provided more and to a certain degree overlapping statistics investigating variance. In case the focus is on a good estimate of the fold change, the researcher should higher account for statistics investigating this measure. Correlation to qRT-PCR or slope and intercept of the regression between qRT-PCR and gene expression fold changes are examples of analyses that could be of higher interest in this context. Focusing on variance, best suited for the data set analysed here are noBg_log_quantile and noBg_log_rsn. Although log_{2}-transformation in combination with quantile normalization has been approved as performing relatively well by Du *et al*. [17] and Dunning *et al*. [8, 10], we decided to make use of robust spline normalization (rsn). In addition to our measures it was selected because rsn is aiming at combining the positive effects of quantile normalization, i.e. preservation of the rank order, and spline interpolation, i.e. continuous mapping of the values, but at the same time circumventing their drawbacks, i.e. discontinuous mapping of intensity values and no rank preservation, respectively [17, 23]. Surprisingly, the use of vst as recommended by Dunning *et al*. [10] and by Du *et al*.[9, 17, 23] and the combination of vst with rsn as successfully used by Du *et al*. [23] did not perform as well as expected. Reasons for this could be the different experimental setups (two replicates per group in the Barnes setup [24] used for validation of vst compared to three to four replicates in our setup) or the use of a newer Illumina chip technology, namely HumanHT-12 v3 chips, in our experiment. vst has been validated based on a pre-released version of the HumanRef-8 v1 Expression BeadChip that contained 19 (25% quantile) to 30 (75% quantile) beads per probe. On the HumanHT-12 v3 chips an average of only 15 beads per probe is available. Since vst makes use of those technical replicates, this could lead to a slightly worse performance on the new chip generation. In general, vst still performs well in stabilizing the variance but is outperformed by noBg_log_quantile, noBg_log_rsn, and noBg_vsn in reflecting the results measured by qRT-PCR. When utilising BeadStudio normalizations, in accordance with Dunning *et al*. [8, 10] who advised against the use of background normalization, we recommend using cubic spline without background normalization (noBg_cubicSpline). As displayed in Figure 1, noBg_cubicSpline outperforms all other BeadStudio normalization methods.

Spike-in or dilution data is frequently used for evaluating different normalization methods [5, 7–10]. If no such data is available for the microarray chip type used, we propose to perform qRT-PCR for genes covering different spectra of expression intensities in order to obtain a measure for judging the quality of pre-processing methods. Thereby, it becomes possible to get an idea of how well different normalization methods are able to reflect the real changes in expression intensities across different expression levels.

In summary, we provide statistical measures based on which researchers can decide on the best suited pre-processing scenario for their own experimental design. If no spike-in data is available, we recommend conducting qRT-PCR for selected, representative transcripts. Thereby, it is possible to estimate the bias of log_{2} ratios obtained from normalized data. In conjunction with the measures for the variability of the data finally the basis for weighing well measured changes versus low and homogeneous variance is delivered and by this means selecting an appropriate normalization method is possible.

## Methods

### Biological experiments

#### Cell culture

HaCaT cells were cultured under standard conditions (REF). Cells were seeded in 96-well (ELISA) or in 24-well (RNA expression profiling) plates and grown overnight to a confluence of approximately 70%. Cells were starved for 3 hours in DMEM containing no FCS and subsequently stimulated with 5 ng/ml of TGF-β1 (R&D Systems) or left unstimulated as controls for 2, 4, and 12 hours.

#### RNA extraction

RNA isolation was carried out using a MagMAX™ Express-96 Magnetic Particle Processor and the MagMAX™-96 Total RNA Isolation Kit according to the manufacturer's protocol. Total RNA concentration was quantified by fluorescence measurement using SYBR Green II (Invitrogen) and a Synergy HT reader (BioTek) as previously described [25]. The RNA quality was characterized by the quotient of the 28 S to 18 S ribosomal RNA electropherogram peak using an Agilent 2100 bioanalyzer and the RNA Nano Chip (Agilent).

#### Amplification, labeling and BeadChip hybridization of RNA samples

Illumina TotalPrep RNA Amplification Kit (Ambion) was used to transcribe 200 ng toRNA according to the manufacture's recommendation. A total of 700 ng of cRNA was hybridized at 58°C for 16 hours to the Illumina HumanHT-12 v3 Expression BeadChips (Illumina). BeadChips were scanned using an Illumina BeadArray Reader and the Bead Scan Software (Illumina). Data is publicly available in ArrayExpress [26] (E-MTAB-265).

#### qRT-PCR

Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) was conducted for eight genes (CDKN1A, CDKN2B, HAND1, JUNB, LINCR, RPTN, SERPINE1, and TSC22D1) known to be deregulated at at least one time point by TGF-β stimulation.

mRNA expression levels of the eight genes were determined by qRT-PCR analysis using a 7900HT Fast Real-Time PCR System (Applied Biosystems) and the Universal ProbeLibrary System (Roche). Gene specific forward and reverse primer sequences were designed using the Universal Probe Library Assay Design Center (Roche). Total RNA was transcribed into cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacture's instructions. qRT-PCR is carried out in a final volume of 12 μl in three replicates for each cDNA sample. Levels of RNA polymerase II were used for normalization of the data. ΔΔCT method was used to relatively quantify mRNA levels of treated samples compared to untreated controls (Additional file 8). Data is publicly available in ArrayExpress [26] (E-MTAB-265).

### Data processing

Data has been processed with BeadStudio version 3.0 and the R Language and Environment for Statistical Computing (R) 2.7.0 [27, 28] in combination with Bioconductor 2.2 [29]. The Bioconductor lumi package [17] has been used for quality control. 25 combinations of background correction, transformation, and normalization methods displayed in Figure 1 were calculated either with methods from BeadStudio, with methods available in the lumi package, or with a combination of BeadStudio and lumi methods.

#### BeadStudio pre-processing

The normalizations executed by Illumina BeadStudio were all applied to the expression values on the original scale. If background adjustment was performed, we used the standard background normalization offered by BeadStudio (bg_*). Cubic Spline, Rank Invariant, and Average methods were used for normalization (for details see BeadStudio Gene Expression Module User Guide [30]). Expression values were then log_{2}-transformed.

#### R pre-processing

*_noNorm data has been log_{2}-transformed using the lumiT() function. Thus, for background corrected data forcePos has automatically been conducted.

forcePositive (forcePos) [17] or rma background adjustment (bgAdjust.affy) [4] available through the lumiB() function of the lumi package were used to transform negative values which can result from BeadStudio background normalization to positive scale to be able to log_{2}-transform the expression values. Background correction referred to as noBg implies that the background normalization has not been applied.

For transforming the data, a simple log_{2}-tranformation (log) or variance-stabilizing transformation (vst) [9] was used.

Data was normalized using quantile normalization (quantile) [31], robust spline normalization (rsn) [17], local regression (loess) [32], or variance stabilization and normalization (vsn) [18]. vst as well as vsn can handle negative values in the data. Thus, neither forcePos nor rma was applied as pre-processing for any of those two methods to not unnecessarily modify the values in artificial ways.

All methods used are implemented in the R packages affy [33], vsn [18], or lumi [17].

### Statistical measures

In the following, the statistical measures used are briefly summarized. Unless otherwise noted, all statistical calculations were performed using R. A small R-package to conduct and reproduce the described analyses is available from the authors upon request. For the visualizations displayed in Figures 3, 6, 7, 8, and Additional files 2, 4, 5, 6, and 7 the smoothScatter() function as implemented within the Bioconductor package geneplotter [34] has been used.

#### Signal to noise ratios

respectively.

Here, k represents, for a given gene, the number of groups, n_{i} the size of group i,
the mean expression level of group i,
the total mean, N the total number of observations, and x_{ij} the j^{th} value in group i. The aim is to maximize
which follows an F-statistic with (k -1; N-k) degrees of freedom. The results for this test are displayed in Figures 2 to 5. For artificial group means
= 6,
= 6, and
= 7, k = 3, n_{1} = n_{2} = n_{3} = 4, and
results to 1.33 and is indicated in Figures 4 and 5 by a grey dashed line. The FDR-corrected [35] p-values for the F-statistic were summarized using their empirical distribution function (Figure 2).

#### Pseudo-ROC curves

One of the main uses of expression arrays is the identification of genes that are differentially expressed under various experimental conditions. A typical identification rule filters genes with p-values and/or fold change exceeding a given threshold. Given a set of known true positives (TP) and false positives (FP), Receiver Operator Characteristic (ROC) curves offer a graphical representation of both specificity and sensitivity for such a detection rule. ROC curves are created by plotting the true positive rate (sensitivity) against false positive rate (1-specificity) obtained at each possible threshold value. Since we only know about TP (20 genes known to be deregulated by TGF-β), we made use of so-called pseudo-ROC curves [22]. The TNs were randomly sampled from the set of transcripts remaining when subtracting the TPs from all transcripts. As a threshold, we used FDR-adjusted p-values [35] of an F-statistic as previously described, this time based on the sample groups for untreated and TGF-β stimulated HaCaT cells at 2 hours.

#### Log_{2} ratios, residual standard deviation, and p-values

The log_{2} ratios, residual standard deviation, and p-values were calculated using linear models in combination with the moderated t-statistic as supplied by limma [21].

#### Regression analysis of fold change values and qRT-PCR measurements

To get an overall impression of how good of a fit of the fold change levels detected using the different normalization methods to the qRT-PCR results are, an orthogonal regression for the observations was performed using the princomp() function as available in the basic R environment.

## Naming conventions

The following naming conventions are used to refer to different normalization methods:

### R normalizations

<background correction>_<transformation>_<normalization>, where:

• <background correction>={bg_forcePos, bg_rma, noBg},

• <transformation>={log, vst},

• <normalization>={loess, quantile, rsn}, and

• <transformation>_<normalization>={vsn}.

### BeadStudio normalizations

<background normalization>_<normalization>, where

• <background normalization>={bg, noBg} and

• <normalization>={cubicSpline, rankInvariant, average}.

## Notes

## Declarations

### Acknowledgements

We would like to thank Simon Anders, Elin Axelsson, Richard Bourgon, Bernd Fischer, Audrey Kauffmann, Gregoire Pau, and Jörn Tödling for helpful discussions and Stephen Gelling for proofreading.

## Authors’ Affiliations

## References

- Schena M, Shalon D, Davis RW, Brown PO: Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science. 1995, 270: 467-470. 10.1126/science.270.5235.467.PubMedView Article
- Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS, Mittmann M, Wang C, Kobayashi M, Horton H: Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol. 1996, 14: 1675-1680. 10.1038/nbt1296-1675.PubMedView Article
- Kuhn K, Baker SC, Chudin E, Lieu MH, Oeser S, Bennett H, Rigault P, Barker D, McDaniel TK, Chee MS: A novel, high-performance random array platform for quantitative gene expression profiling. Genome Res. 2004, 14: 2347-2356. 10.1101/gr.2739104.PubMed CentralPubMedView Article
- Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP: Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003, 4: 249-264. 10.1093/biostatistics/4.2.249.PubMedView Article
- Cope LM, Irizarry RA, Jaffee HA, Wu Z, Speed TP: A benchmark for Affymetrix GeneChip expression measures. Bioinformatics. 2004, 20: 323-331. 10.1093/bioinformatics/btg410.PubMedView Article
- Lim WK, Wang K, Lefebvre C, Califano A: Comparative analysis of microarray normalization procedures: effects on reverse engineering gene networks. Bioinformatics. 2007, 23: i282-i288. 10.1093/bioinformatics/btm201.PubMedView Article
- Jiang N, Leach LJ, Hu X, Potokina E, Jia T, Druka A, Waugh R, Kearsey MJ, Luo ZW: Methods for evaluating gene expression from Affymetrix microarray datasets. BMC Bioinformatics. 2008, 9: 284-10.1186/1471-2105-9-284.PubMed CentralPubMedView Article
- Dunning MJ, Barbosa-Morais NL, Lynch AG, Tavare S, Ritchie ME: Statistical issues in the analysis of Illumina data. BMC Bioinformatics. 2008, 9: 85-10.1186/1471-2105-9-85.PubMed CentralPubMedView Article
- Lin SM, Du P, Huber W, Kibbe WA: Model-based variance-stabilizing transformation for Illumina microarray data. Nucleic Acids Res. 2008, 36: e11-10.1093/nar/gkm1075.PubMed CentralPubMedView Article
- Dunning MJ, Ritchie ME, Barbosa-Morais NL, Tavare S, Lynch AG: Spike-in validation of an Illumina-specific variance-stabilizing transformation. BMC Res Notes. 2008, 1: 18-10.1186/1756-0500-1-18.PubMed CentralPubMedView Article
- Ding LH, Xie Y, Park S, Xiao G, Story MD: Enhanced identification and biological validation of differential gene expression via Illumina whole-genome expression arrays through the use of the model-based background correction methodology. Nucleic Acids Res. 2008, 36: e58-10.1093/nar/gkn234.PubMed CentralPubMedView Article
- Xie Y, Wang X, Story M: Statistical methods of background correction for Illumina BeadArray data. Bioinformatics. 2009, 25: 751-757. 10.1093/bioinformatics/btp040.PubMed CentralPubMedView Article
- McCall MN, Irizarry RA: Consolidated strategy for the analysis of microarray spike-in data. Nucleic Acids Res. 2008, 36: e108-10.1093/nar/gkn430.PubMed CentralPubMedView Article
- Wilder SP, Kaisaki PJ, Argoud K, Salhan A, Ragoussis J, Bihoreau MT, Gauguier D: Comparative analysis of methods for gene transcription profiling data derived from different microarray technologies in rat and mouse models of diabetes. BMC Genomics. 2009, 10: 63-10.1186/1471-2164-10-63.PubMed CentralPubMedView Article
- HumanHT-12 v3 Expression BeadChip. [http://www.illumina.com/Documents/products/datasheets/datasheet_humanht_12.pdf]
- Illumina
^{®}BeadStudio. [http://www.illumina.com/Documents/products/datasheets/datasheet_beadstudio.pdf] - Du P, Kibbe WA, Lin SM: lumi: a pipeline for processing Illumina microarray. Bioinformatics. 2008, 24: 1547-1548. 10.1093/bioinformatics/btn224.PubMedView Article
- Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M: Variance stabilization applied to microarray data calibration and to the quantification of differential expression. Bioinformatics. 2002, 18 (Suppl 1): S96-104.PubMedView Article
- Scheffé H: The Analysis of Variance. 1959, New York, Wiley, John & Sons, Inc
- Cui X, Churchill GA: Statistical tests for differential expression in cDNA microarray experiments. Genome Biol. 2003, 4: 210-10.1186/gb-2003-4-4-210.PubMed CentralPubMedView Article
- Smyth GK: Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol. 2004, 3:
- Bourgon RW: Chromatin immunoprecipitation and high-density tiling microarrays: a generative model, methods for analysis, and methodology assessment in the absence of a "gold standard". PhD Thesis. 2006, University of California, Berkeley, Department of Statistics
- Using lumi, a package processing Illumina Microarray. [http://www.bioconductor.org/packages/2.2/bioc/html/lumi.html]
- Barnes M, Freudenberg J, Thompson S, Aronow B, Pavlidis P: Experimental comparison and cross-validation of the Affymetrix and Illumina gene expression analysis platforms. Nucleic Acids Res. 2005, 33: 5914-5923. 10.1093/nar/gki890.PubMed CentralPubMedView Article
- Schmidt DM, Ernst JD: A fluorometric assay for the quantification of RNA in solution with nanogram sensitivity. Anal Biochem. 1995, 232: 144-146. 10.1006/abio.1995.0025.PubMedView Article
- Parkinson H, Kapushesky M, Kolesnikov N, Rustici G, Shojatalab M, Abeygunawardena N, Berube H, Dylag M, Emam I, Farne A: ArrayExpress update--from an archive of functional genomics experiments to the atlas of gene expression. Nucleic Acids Res. 2009, 37: D868-D872. 10.1093/nar/gkn889.PubMed CentralPubMedView Article
- R Development Core Team: R: A Language and Environment for Statistical Computing. 2008, Vienna, Austria, R Foundation for Statistical Computing
- Gentleman RC, Ihaka R: R. A Language for Data Analysis and Graphics. Journal of Computational and Graphical Statistics. 1996, 5: 299-314. 10.2307/1390807.
- Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J: Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004, 5: R80-10.1186/gb-2004-5-10-r80.PubMed CentralPubMedView Article
- Illumina Inc. [http://www.illumina.com/]
- Bolstad BM, Irizarry RA, Astrand M, Speed TP: A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics. 2003, 19: 185-193. 10.1093/bioinformatics/19.2.185.PubMedView Article
- Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, Speed TP: Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res. 2002, 30: e15-10.1093/nar/30.4.e15.PubMed CentralPubMedView Article
- Gautier L, Cope L, Bolstad BM, Irizarry RA: affy--analysis of Affymetrix GeneChip data at the probe level. Bioinformatics. 2004, 20: 307-315. 10.1093/bioinformatics/btg405.PubMedView Article
- geneplotter: Graphics related functions for Bioconductor. [http://www.bioconductor.org/packages/2.2/bioc/html/geneplotter.html]
- Benjamini Y, Hochberg Y: Controlling the false discovery rate: A practiacal and powerful approach to multiple testing. J Roy Statist Soc Ser B. 1995, 57: 289-300.
- gplots: Various R programming tools for plotting data. [http://cran.r-project.org/web/packages/gplots/index.html]
- Livak KJ, Schmittgen TD: Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001, 25: 402-408. 10.1006/meth.2001.1262.PubMedView Article

## Copyright

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.