The limited overlap of our gene expression and protein data sets warrants discussion. Cellular RNA and protein levels are imperfectly associated, with a correlation coefficent <0.7,31 likely owing to a combination of post-transcriptional effects, rapid translation and protein turnover. In addition, our SILAC methods were unlikely to detect secreted proteins, such as RELN and other extracellular matrix proteins, that were significantly perturbed in our microarrays. In fact, of the 483 significantly misexpressed genes identified by microarray analysis, most were poorly detected as proteins by mass spectrometry (Supplementary Table 9). Of those significantly misexpressed genes detected and identified as significantly perturbed in three of our four SILAC mass spectrometry analyses, there was an enrichment in the predicted direction of change (binomial test P-values: 5.6053e–04, 3.9675e–05, 0.1442, 2.0013e–09), while there was no effect in the opposite direction. Our gene and protein data sets were more consistent at the pathway level (Supplementary Figure 5D).
