To investigate the presence of local associations in addition to the most significant SNP reported for each probe, we carried out a conditional analysis with imputed genotypes. To this end, the cis-eQTL analysis was repeated for all gene expression probes that had produced an association with a P value of no more than 5 × 10−8, conditioning on the initial association. If no additional independent eQTL are present, it is expected that this will not identify any additional significant associations. Conversely, any associations that are independent of the first one are expected to remain and may increase in significance. This process was iterated until no further significant associations remained (P < 5 × 10−8). To confirm the presence of an independent signal, this process was repeated for the second and third most significantly associated SNP for a given probe. Only examples where the secondary signal was present in all instances are reported.
