QIMR cohorts were drawn from larger batches genotyped over an extended period using several different Illumina genotyping microarrays. The microarrays used were (1) Global Screening Array v1 or v2 used for AGDS and GBP, and for TWINS participants either GSA (N = 48); (2) Illumina Omni or Core+Exome family chips (Core+Exome N = 1,023, PsychArray N = 255, OmniExpress N = 102 and 2.5M N = 321; total N = 1,701) or (3) older Illumina HapMap-derived chips (370K N = 3,728, 610K N = 2,319, 317K N = 580 and 660K N = 27; total N = 6,654). Per-batch imputation QC removed variants with GenTrain score <0.6, MAF <0.01, SNP call rate <95% and HWE deviation (P < 1 × 10−6). Genotypes from each of the three Illumina microarray families were merged for the core set of markers that passed QC in all batches, then were imputed using the TOPMed Imputation Server with the TOPMed-r2 reference panel64,71. The core set used ∼441K, ∼232K and ∼280K markers for (1), (2) and (3), respectively. Association analysis was performed using SAIGE with the
