A score of 1f means "eQTL + TF binding / DNase peak", whereas a score of 1d means "eQTL + TF binding + any motif + DNase peak", and 1b means "eQTL + TF binding + any motif + DNase Footprint + DNase peak".\cite{help} 1a is the "best" possible score, in that it corresponds to he highest chance of functionality, followed in alphabetical order by the other "1" values (letters a through f). Of the 2,338 SNPs tested, only 2 (0.08%) had the highest score of 1a: namely,
Of these, the first is an intergenic variant that has only been reported in one study--namely, the one by Davies et al. (2018) included here.\cite{catalog}\cite{Davies_2018} The second is an intron that also has only ever been reported by Davies et al. (2018).\cite{catalog}\cite{Davies_2018}
One can also only consider SNPs for which a Regulome score could be obtained at all (n = 1,796). Of these, 616 (34%) had scores of 6, the most common score, whereas 2 (0.11%) had scores of 1a, both the "best" and the least common.
Hill et al.\cite{Hill2019} also analyzed SNP hits for intelligence using Regulome. Their results, illustrated in Figure 1C in their paper, showed the score of 7 to be the most common, including a total of 13.8% of all of their SNPs. This is a rather higher percentage than what I found, though still much lower than reported by Savage et al. (2018). I also had a much higher % of SNPs with scores of 6 (26.3%) compared to Hill et al. (12.6%), and Hill et al. also found a weirdly high % of SNPs with "perfect" scores of 1a (4.5%), much higher than I found (0.09%). The percent distributions of Regulome scores in the current paper is compared to those from Savage et al. and from Hill et al. in Figure 2 below.