|
|
Post by jhuyghe on May 20, 2016 21:53:06 GMT
Hello Jian,
How can I calculate standard errors of the row and column sum estimates similar to what was done in Supplementary Table 3 of your Nature Genetics paper describing GREML-LDMS? Is there a way to get the covariance matrix of the genetic components from GCTA?
Thanks!
Jeroen
|
|
|
|
Post by Jian Yang on May 24, 2016 8:09:38 GMT
The SE of the estimate of each variance component is reported in the *.hsq file.
Here is an example of a model of fitting 7 variance components. The third last row shows the sum of estimates and its SE.
Source Variance SE
V(G1) 0.021034 0.010191
V(G2) 0.002085 0.015312
V(G3) 0.038612 0.009742
V(G4) 0.059068 0.007307
V(G5) 0.038531 0.006868
V(G6) 0.042273 0.006732
V(G7) 0.044826 0.006129
V(e) 0.755290 0.018957
Vp 1.001719 0.006969
V(G1)/Vp 0.020998 0.010171
V(G2)/Vp 0.002081 0.015286
V(G3)/Vp 0.038546 0.009715
V(G4)/Vp 0.058967 0.007254
V(G5)/Vp 0.038465 0.006838
V(G6)/Vp 0.042200 0.006697
V(G7)/Vp 0.044749 0.006087
Sum of V(G)/Vp 0.246006 0.018741
logL -21285.492
n 43366
The variance covariance matrix of the estimates are available in the log output. Here is an example.
Sampling variance/covariance of the estimates of variance components:
0.000103864 -2.25528e-05 -3.91735e-07 -1.65974e-07 -6.41693e-07 1.50848e-07 2.15811e-07 -7.87054e-05
-2.25528e-05 0.000234463 -2.26976e-05 -2.28322e-06 -2.00219e-06 -1.03407e-06 -1.38185e-06 -0.000182865
-3.91735e-07 -2.26976e-05 9.49064e-05 -1.38152e-05 -4.1499e-06 -2.46442e-06 -3.07761e-06 -4.47772e-05
-1.65974e-07 -2.28322e-06 -1.38152e-05 5.33875e-05 -1.16231e-05 -4.30146e-06 -3.15843e-06 -1.1688e-05
-6.41693e-07 -2.00219e-06 -4.1499e-06 -1.16231e-05 4.71741e-05 -1.27284e-05 -4.94644e-06 -6.84187e-06
1.50848e-07 -1.03407e-06 -2.46442e-06 -4.30146e-06 -1.27284e-05 4.5319e-05 -1.42257e-05 -5.95413e-06
2.15811e-07 -1.38185e-06 -3.07761e-06 -3.15843e-06 -4.94644e-06 -1.42257e-05 3.75654e-05 -5.63714e-06
-7.87054e-05 -0.000182865 -4.47772e-05 -1.1688e-05 -6.84187e-06 -5.95413e-06 -5.63714e-06 0.000359377
|
|
|
|
Post by jhuyghe on May 26, 2016 21:27:20 GMT
Thanks! I missed that the covariances where in the log output.
Jeroen
|
|
