So I aimed to count how many SNPs, and what percent of total SNPs, that were reported in the included studies were classified into each category.
Previous work focusing on SNPs associated with intelligence at a genome-wide significant level has already been done. For instance, Sniekers et al. (2017) found 336 such SNPs, of which 162 (48%) were introns.
I used the Regulome online tool to test for the probability that each SNP had a regulatory function, i.e. that it actually might have a causal effect on any phenotype.\cite{Boyle2012} Other previous GWASs and GWAS meta-analyses focusing on intelligence have also used Regulome for this purpose (e.g. \cite{Hill2019},\cite{Savage_2018}).
Testing for the presence of a "winner's curse"