The fraction of replicated observations between this and the former study was between 16 and 20%, which included replications by LD. As these represent eQTLs replicated in two different studies, they should be highly reliable, justifying in-depth analysis without further validation. However, the large discrepancy between the two studies was unexpected. Likely reasons include technical differences between the two studies, in particular the different gene expression profiling platforms applied in our (Illumina) and the former study (Agilent Technologies), which used different probes for most genes. This can lead to different expression data, given the common occurrence of variant transcripts, SNP-interference with probe hybridization and inconsistent gene annotation.45, 46 Additional technical differences concern the genotyping arrays and differences in data preprocessing.
