(detailed sample information can be found in Supplementary Table 1 online). All samples from CTX, CN and CB were analyzed using Affymetrix U133A microarrays, whereas samples from CTX_95 were analyzed using Affymetrix U95A/v2 microarrays. To ensure the highest possible level of data quality, rigorous quality control procedures were implemented to eliminate mis-targeted and nonspecific probes on the microarrays before generating expression values23, identify and remove outlier samples from the data sets, and carry out additional normalization to remove ‘batch effects’ introduced by combining data from multiple studies24 (Methods).