
Post by Asling_GCTA on Jul 30, 2014 8:39:02 GMT
Hello,
I use GCTA to estimate the variance explained by GRM from a subset of snps for a binary trait. (0/1).Here is the reml result:
Source Variance SE V(G) 0.003794 0.003079 V(e) 0.132379 0.003814 Vp 0.136173 0.002400 V(G)/Vp 0.027865 0.022593 The estimate of variance explained on the observed scale is transformed to that on the underlying scale: (Proportion of cases in the sample = 0.174438; Userspecified disease prevalence = 0.130000) V(G)/Vp_L 0.055307 0.044844 logL 3190.744 logL0 3189.958 LRT 1.573 df 1 Pval 0.1049 n 6455
It seems that although GCTA gave the estimated variances on the liability scale, the Pvalue of LRT is not significant. So my question is that does this mean that the estimate itself is not valid?
Thank you!
Asling



Post by Zhihong Zhu on Jul 30, 2014 14:07:36 GMT
Hi Asling,
Yes, we are unable to conclude that the h^2 explained by the subset of SNPs is different from zero.
Cheers, Zhihong



Post by Asling_GCTA on Jul 31, 2014 8:34:48 GMT
Hi Asling, Yes, we are unable to conclude that the h^2 explained by the subset of SNPs is different from zero. Cheers, Zhihong Thank you!

