Anne-Marie Koop

54 Gene 1.1 ST arrays (Affymetrics) were performed according to standard Affymetrix protocols. Quality control and normalization Scans of the Affymetrix arrays were processed in the MADMAX pipeline (Nutrigenomics Consortium, Wageningen, The Netherlands) 9 using Bioconductor software packages. Quality control was carried out by visual inspection of the heat map, Affymetrics Quality Control metrics, Relative Log Expression-plot, Normalized Unscaled Standard Error-plot and hierarchical clustering. Expression levels of probe sets were normalized using the robust multi-array average algorithm 10 with 19239 transcripts passing the filter. Probe sets were assigned to genes using the custom CDF library version 15.1.1. Array data are deposited at the Gene Expression Omnibus (GEO) database (GSE46863). Differential expression of individual genes Differentially expressed probe sets were identified using an IBMT regularized t-test. 11 P values were corrected for multiple testing using a false discovery rate method. Probe sets that satisfied the criterion of a false discovery rate <1% were considered significantly regulated. Gene set enrichment analysis Gene set enrichment analysis (GSEA, version 3.1) 12 was used to explore changes in the global gene expression pattern. Out of 899 predefined gene sets (gene set size set to min=50, max=500), those passing the criteria false discovery rate (FDR) <15%, nominal P value <0.05 and normalized enrichment score >1.3 were considered significant. All gene sets available were obtained from the C2-curated Molecular Signatures Database. DAVID Database for Annotation, Visualization and Integrated Discovery (DAVID) software was used to categorize genes into biological processes. 13,14 In DAVID, statistical significance of differential expression of a biological process was assessed using moderated  t -tests; p-values were adjusted for multiple testing to control false discovery rate using the Benjamini method. P<0.01 was considered significant.