Hi,
I try to estimate LDSC h2 by hand similar to Figure 2 in "LD Score regression distinguishes confounding from polygenicity in genome-wide association studies". I work with a continuous trait with a GCTA h2 of ~0.20 (P<0.0001).
So, far I just created LDSC bins according to quantiles (e.g. 100), generated LD scores means and Chi square means per bin, and plotted them. (I have not even started using gls). However, my LDSC h2 estimates are not even near the derived h2 using the LDSC python software.
LDSC estimates:Intercept(SE) : 0.992(0.0067)
h2 (SE) : 0.26(0.09)
OLS R estimates:Call:
lm(formula = chi2 ~ L2, data = L2.chi)
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.9906202 0.0049205 201.326 < 2e-16 ***
L2 0.0013436 0.0001636 8.215 8.8e-13 ***
Any chance you could point me in the right direction?
Is it a question of the regression weights, which I have not included yet?
Many thanks,
Beate