|
|
Post by ktansey on Jun 19, 2014 8:35:34 GMT
I'm attempting to run a bivariate analysis where one phenotype is case-control and the other is a quantitative trait, which is now possible in the latest version of GCTA. I am a slightly confused about the parameter options for this analysis. Do I need to set a prevalence with --reml-bivar-prevalence and if so, is it okay to just have one prevalence rate for the case control sample?
I have assumed this is the case and I run the analysis and I get a negative value for (rG). The interpretation of this result is not clear to me and I am wondering if there is a command line error or there is genuine negative genetic correlation (risk locus for one trait is protective for another)?
Thanks
Katherine
|
|
|
|
Post by Jian Yang on Jun 19, 2014 10:53:06 GMT
If you want to transform the estimate on the observe scale to that on the liability scale for the disease, you need to set a prevalence. Yes, just one for the disease.
A significant negative genetic correlation means genes have negative effects on the trait could increase the risk to the disease.
|
|
ella
New Member
Posts: 1 
|
Post by ella on Feb 13, 2017 12:22:05 GMT
I have completed a bivariate analysis for a binary and quantitative trait and, as suggested here, set the prevalance only for the binary trait, however, I don't fully understand the output (below). Should I be comparing the values from the observed scale for the quantitative trait with the liability scale for the binary trait? And can I just ask for clarification on the meaning of the line stating "prevalence=-1.000000 for disease 2"?
V(G1)/Vp_tr1 0.046371 0.019922
V(G1)/Vp_tr2 0.043632 0.020446
V(G2)/Vp_tr1 0.182138 0.025965
V(G2)/Vp_tr2 0.230915 0.026984
V(G3)/Vp_tr1 0.075899 0.027936
V(G3)/Vp_tr2 0.140351 0.027117
The estimate of variance explained on the observed scale is transformed to that on the underlying scale:
(Proportion of cases in the sample = 0.138904; User-specified disease prevalence = 0.130000 for disease 1 and = -1.000000 for disease 2)
V(G1)/Vp_tr1_L 0.110815 0.047608
V(G1)/Vp_tr2_L inf inf
V(G2)/Vp_tr1_L 0.435266 0.062051
V(G2)/Vp_tr2_L inf inf
V(G3)/Vp_tr1_L 0.181380 0.066760
V(G3)/Vp_tr2_L inf inf
Thanks very much,
Ella
|
|
