From these normalised entity lists, a significance cut off of p ≤ 0.01 and fold change cut off of ≥ 2 were applied.
To correct for any bias due to the separation of pre- and post-treatment samples on unique TMAs, AQUA results for the matched tissue samples on each TMA were median normalised prior to analysis of significance by the Wilcoxon matched-pairs test.
The gene expression data were normalised using quantile normalisation within the bead array package (Dunning et al, 2007) and differential gene expression was assessed using significance analysis of microarrays (SAM; Tusher et al, 2001) using the siggenes package.
We have assessed seven methods used across the literature for detecting differential expression from spectral count quantification: Student's t-test, significance analysis of microarrays (SAM), normalised spectral abundance factor (NSAF), normalised spectral abundance factor-power law global error model (NSAF-PLGEM), spectral index (SpI), DESeq and QSpec.
To test significance of differences in normalised peak areas between different strains, a two-tailed Student's T-test was applied.
To confirm statistical significance for cluster sets, normalised values were exported and analysed by GraphPad Prism 4. All gene identities were confirmed by BLAST search of the probeset sequences using the Entrez nucleotide databases [ 36].
First, the change in level of expression of equivalent spots had to reach significance using the standard deviation of the normalised spot volumes and also a Student's t-test comparing Gefitinib-treated and DMSO-treated gels.
For normalised pan GR mRNA, diagnostic group differences in expression approached significance [ANCOVA F 2, 89) = 2.74, p = 0.07, co-varying for RIN and brain pH; Figure 2A].
Normalised miRNAs with <20% missing values were included for Significance Analysis of Microarrays (SAM) (http://www-stat.stanford.edu/Btibs/SAM/) and clustering analysis.
Thus, to correctly evaluate the statistical significance of the remaining components of variation, the normalised data matrix needs to be adjusted by removing this background variation, by subtracting out the top principal component.
