This catalog is described in the accompanying paper 'Conditional eQTL Analysis Reveals Allelic Heterogeneity of Gene Expression' by Rick Jansen et al (link to paper).
The data for this study was measured by two Dutch Biobanks: NESDA ( www.nesda.nl ) and NTR ( www.tweelingenregister.org ).

Abstract

Gene expression is often strongly associated with multiple single nucleotide polymorphisms (SNPs), which may reflect one or multiple independent associations. Conditional eQTL (expression quantitative trait loci) analysis allows to distinguish between dependent and independent eQTLs, which is not possible in the currently available eQTL databases. We performed conditional eQTL analysis using 4,896 peripheral blood gene expression samples. Our analysis showed that 35% of genes with a cis eQTL have at least two independent cis eQTLs; for several genes up to 13 independent cis eQTLs were identified. Also, 12% (671) of the independent cis eQTLs identified in conditional analyses were not significant in unconditional analyses. The number of GWAS catalog SNPs identified as eQTL in the conditional analyses increases with 24% as compared to unconditional analyses. We provide a online conditional cis eQTL mapping catalog for whole blood, which can be used to lookup eQTLs more accurately than in standard unconditional whole blood eQTL databases.

Methods (in short)

eQTL effects were defined as cis when probeset–SNP pairs were at distance < 1M base pairs (Mb), and as trans when the SNP and the probe set were separated by more than 5 Mb on the genome according to hg19. For each probe set that displayed a statistically significant association with at least one SNP in the cis region, we identified the most significantly associated SNP (E1 SNP). Conditional eQTL analysis was carried out by first residualizing probeset expression using the corresponding E1 SNP and then repeating the eQTL analysis using the residualized data. Then, for each probe set the most significant SNP was selected (E2 SNP) and each probeset was residualized using the E1 and E2 SNPs, and eQTL analysis was repeated using the residualized expression. This was repeated until no more significant associations were found between residualized expression and SNP data (after up to 12 rounds of conditional analysis). Of note, the eQTL P-values reported in this website are based on the complete sample with related subject and thus are too liberal: however the FDR takes into account the family structure and should be used to draw conclusions.