Table 3. Number of cis eQTLs identified in previous studies and replicated in GHS.
Stranger et al. | Dixon et al. | Schadt et al. | ||||
Level of signifi-cance | Number of eQTLs at level of significance | Percent signifi-cant in GHS* | Number of eQTLs at level of significance | Percent significant in GHS* | Number of eQTLs at level of significance | Percent signifi-cant in GHS* |
>10−8 | 86 | 55.8 | 110 | 50.9 | 928 | 47.9 |
10−8–10−10 | 63 | 69.8 | 162 | 50.0 | 168 | 57.7 |
10−10–10−15 | 144 | 63.2 | 237 | 54.8 | 211 | 57.3 |
10−15–10−20 | 60 | 70.0 | 102 | 60.7 | 120 | 66.7 |
10−20–10−25 | 38 | 89.5 | 73 | 65.7 | 73 | 67.1 |
≤10−25 | 48 | 70.8 | 89 | 67.4 | 103 | 73.8 |
All | 439 | 66.7 | 773 | 56.5 | 1603 | 54.1 |
* Comparisons were based on sets of gene expressions overlapping between each study and GHS and were restricted to autosomal cis eQTLs. All cis eQTLs considered significant in each study were retrieved and replication was assessed in GHS (P<3.9×10−6 correcting for 12,808 gene expressions).
For Stranger et al [1], data were extracted from Table S2. We considered as significant the associations found in at least 3 HAPMAP populations. For Dixon et al [2], data were extracted from Table S1 and trans eQTLs were excluded. Matching of probes was done using a table provided by the authors on their web site. For Schadt et al [3], cis eQTLs considered significant (First.Pass.Indicator set to 1) were extracted from Table S3. For each eQTL, we selected in GHS the P-value of the best cis eSNP. The full data used to generate this table are provided in Files S2–S4.