Table 2. Predicted and observed number of genome-wide-significant hits and PGS R², for large-scale GWAS efforts to date for height, BMI, EduYears, and self-rated health, assuming 250k effective SNPs (i.e., independent haplotype blocks) of which 20k trait-affecting, using averaged GREML estimates from Table 1 for setting SNP heritability and CGR.

Notes on the sources for the large-scale GWAS efforts are listed in S3 Table.

Phenotype	Main studies			Architecture		Number of hits				PGS R2 using all SNPs
	Study	N	C **	$h_{\text{SNP}}^2$	CGR	Study	Theory|CGR		Attenuation*	Study	Theory|CGR		Attenuation*
							<1	=1			<1	=1
Height	Wood et al. (2014) [1]	253,288	79	44.9%	0.965	697	647.26	700.24	8%	13.5%	13.2%	14.0%	6%
	Allen et al. (2010) [46]	183,727	61	44.9%	0.965	180	292.03	320.77	9%	10.0%	10.5%	11.1%	6%
	Weedon et al. (2008) [47]	13,665	5	44.9%	0.965	7	0.00	0.00	n.a.	2.9%	1.0%	1.1%	7%
BMI	Locke et al. (2015) [2]	339,224	125	21.9%	0.917	97	188.52	241.07	22%	6.5%	4.3%	5.0%	14%
	Speliotes et al. (2010) [48]	123,865	46	21.9%	0.917	19	5.48	7.64	28%	2.5%	1.8%	2.1%	15%
	Willer et al. (2008) [49]	32,387	15	21.9%	0.917	1	0.01	0.02	65%	n.a.	0.5%	0.6%	16%
EduYears	Okbay et al. (2016) [7]	405,072	65	18.2%	0.783	162	115.28	235.90	51%	n.a.	2.7%	4.1%	36%
	Okbay et al. (2016) [7]	293,723	64	18.2%	0.783	74	39.30	88.93	56%	3.9%	2.0%	3.2%	36%
	Rietveld et al. (2013) [50]	101,069	42	18.2%	0.783	1	0.63	1.64	62%	2.5%	0.8%	1.2%	38%
Self-rated health	Harris et al. (2016) [51]	111,749	1	15.7%	0.468	13	1.35	1.35	0%	n.a.	0.2%	1.0%	78%

* Attenuation measures the relatively loss in expected power and R² due to a CGR in accordance with averaged GREML estimates from Table 1.

** C denotes the number of studies in the meta-analysis.