Table 1.
Overview of the main findings from the various stages of this study
| Analysis | Objective | Number of tests undertaken | Number of results surviving multiple testing |
|---|---|---|---|
| Two-sample MR (CpG -> complex trait) | Identify potential CpG sites where DNA methylation may mediate the influence of genetic variants on complex traits | 4 215 592 | 1148 |
| Joint likelihood mapping | Assess the likelihood that results from the previous analysis are observed due to two separate causal variants which are in LD with one and other | 1148 | 348 |
| Reverse two-sample MR (complex trait -> CpG) | Evaluate potential evidence for reverse causation, i.e. complex trait influences DNA methylation levels | 348 | 2a |
| Replication two-sample MR | Validate results for 14 complex traits using data from the BIOS QTL browser and UK Biobank study | 128 | 101 |
| Two-sample MR (CpG -> gene expression) | Investigate whether meQTL used as instruments in the initial analysis overlap with variants known to influence nearby gene expression (i.e. whether they are also cis-eQTL) | 348 | 306 |
a
Both effects observed in the reverse MR analysis were based on a single genetic instrument, similar to findings in the initial MR at these loci (CpG -> complex trait). We are therefore unable to robustly distinguish the direction of effect between methylation and complex trait for these associations.