|
Post by clement on Apr 2, 2014 16:27:25 GMT
Hello Jian,
I adjusted my outcome variable by age, sex, and covariates other than PCs. Then I ran GCTA with PCs as qcovar. I got the following output:
Source Variance SE V(G) 0.989846 0.452480 V(e) 0.000001 0.444583 Vp 0.989847 0.055956 V(G)/Vp 0.999999 0.449143 logL -313.036 logL0 -316.718 LRT 7.364 df 1 Pval 0.003 n 632
Is it safe to say that a significant portion of the variance in the score is explained by the SNPs in the GWAS, but the sample size is too small to get a precise estimate? Also, the V(G)/Vp is 0.999999 whenever I included age/sex/etc (either by adjusting the score before running GCTA or including them as covariates during GCTA).
Thank you. Clement
|
|
|
Post by Zhihong Zhu on Apr 3, 2014 0:20:15 GMT
Hi Clement,
I'm afraid not, due to the large SE. The confident interval could be (0.1, 1), (mean - 2*SE, mean + 2*SE) .
Cheers, Zhihong
|
|
|
Post by Jian Yang on Apr 3, 2014 1:16:11 GMT
Hello Jian, I adjusted my outcome variable by age, sex, and covariates other than PCs. Then I ran GCTA with PCs as qcovar. I got the following output: Source Variance SE V(G) 0.989846 0.452480 V(e) 0.000001 0.444583 Vp 0.989847 0.055956 V(G)/Vp 0.999999 0.449143 logL -313.036 logL0 -316.718 LRT 7.364 df 1 Pval 0.003 n 632 Is it safe to say that a significant portion of the variance in the score is explained by the SNPs in the GWAS, but the sample size is too small to get a precise estimate? Also, the V(G)/Vp is 0.999999 whenever I included age/sex/etc (either by adjusting the score before running GCTA or including them as covariates during GCTA). Thank you. Clement Yes, the sample size is too small to get a precise estimate although the LRT is significant.
|
|