Pancreatic Adenocarcinoma vs. Healthy control

Data pre-processing was performed using Bioconductor packages within the R statistical environment. The arrayMvout package was utilised to implement stringent quality control criteria on the datasets. Nine quantitative features of array quality, computed on each array, were evaluated: ABG (average background); SF (scale factor); PP (percent of present calls); AR (actin 30/50 ratio); GR (GAPDH 30/50 ratio); median normalised unscaled standard error (NUSE); median relative log expression (RLE); RLE-IQR (interquartile range of IQR per array, to measure variability in RLE) and RNAS (slope of RNA degradation measure). All data files passing the quality control checks were subsequently jointly normalized to create a global space of gene expression in pancreatic cancer. Differential expression analysis between biological groups was performed using limma. Duplicate correlation method from limma was used to adjust for replicates. The Benjamini and Hochberg (BH) false discovery rate (FDR) was used for multiple testing corrections.