We estimated LD Scores from the European ancestry samples in the 1000 Genomes Project7 (EUR) using an unbiased estimator8 of r2 with 1 centiMorgan (cM) windows, singletons excluded (MAF > 0.13%) and no r2 cutoff. Standard errors were estimated by jackknifing over blocks of individuals, and we used these standard errors to correct for attenuation bias in LD Score regression (i.e., the downward bias in the magnitude of the regression slope that results when the regressor is measured noisily, see Online Methods).
