Each input GWAS is conducted by meta-analysing the relevant input files in MTAG74. All analyses are conducted allowing for sample overlap and setting all genetic correlations equal to unity. However, we allow the SNP-heritability parameter to vary across input files. Even though MTAG produces a separate output file for each input file, the assumption of perfect genetic correlation ensures that the SNP coefficients in each output file are a constant multiple of each other (hence the PGIs generated by the output files are the same). In all analyses that follow, we adopt the convention of designating the output file with the highest estimated SNP heritability as the input GWAS (this matters for the expected R2 calculation but nothing else). The details of the heritability estimation are described below, in the subsection “Criterion for Inclusion in Repository” in Section III.
