jayce
New Member
Posts: 1
|
Post by jayce on May 27, 2016 8:49:03 GMT
hi all
i encountered a problem when i run --reml that it stopped half-way and threw me an error( version gcta_1.25.3): Error: analysis stopped because more than half of the variance components are constrained. The result would be unreliable.
Please have a try to add the option --reml-no-constrain.
if i added -no-constrain in --reml, then my results of V(e) is 0.00 for older version (gcta_1.22): Source Variance SE
V(G) 0.342622 0.065586
V(e) -0.108123 0.062106
Vp 0.234499 0.009811
V(G)/Vp 1.461080 0.259990
The estimate of variance explained on the observed scale is transformed to that on the underlying scale:
(Proportion of cases in the sample = 0.578689; User-specified disease prevalence = 0.010000)
V(G)/Vp_L 0.826863 0.147135
logL 282.806
logL0 267.915
LRT 29.783
df 1
Pval 2.416e-08
n 1220
for latest version gcta_1.25.3 (added -no-constrain), it throws me this error: Error: the information matrix is not invertible.
My understanding of V(e) is referring variance error, but can it also referring to Environments variants (error), may if the understanding is correct?
Thanks very much for your help..
~j
|
|
|
Post by Jian Yang on Jun 3, 2016 5:03:40 GMT
This suggests that estimate is not reliable. The previous version works because it uses an approximate approach to invert the matrix. The latest version is more rigorous in checking the singularity of the matrix. You might need more samples.
|
|