For each of the 47 phenotypes for which we constructed a single-trait and/or multi-trait PGI, Table 1 lists the total sample size included in the GWAS summary statistics (Total N), followed by the sample-size contributions from three separate sources. For comparison, we also report the sample size of the largest GWAS whose summary statistics are in the public domain (Public N). With three exceptions, Total N exceeds Public N. Two exceptions are height and BMI, where our UKB sample inclusion filters lead to a slightly smaller sample size than the Public N. The remaining exception is cognitive performance, where the sample size of our GWAS is smaller due to overlap between the discovery sample in the largest GWAS with publicly available summary statistics and some of our Repository cohorts. For the remaining phenotypes, the gains in sample size relative to the public N are often substantial, and driven by our inclusion of summary statistics from large-scale GWASs conducted in 23andMe, UKB, or both. Table 1 also shows the 36 and 35 phenotypes for which we created single-trait and multi-trait PGIs, respectively.
