A novel locus was declared if the lead SNV reached exome-wide significance in the combined meta-analysis of discovery and replication data (P < 2.12 × 10−7) and replicated with Bonferroni-adjusted significance (P < 0.0042 for 12 tests) in the replication data alone. In addition, the directions of effect between the discovery and replication data were required to be concordant, taking into account the inverse relationship between RR from our discovery data and HR from the replication data.
