|
Post by carolynjumper on May 29, 2014 8:00:18 GMT
I have estimated the total heritability of a group of individuals in a case-control study, and I now I would like to estimate the heritability of a particular subset of those individuals. Is there a way to estimate the heritability and its standard error for the subset without restricting the GRM to include only the subset of individuals?
I assume that using the GRM for all individuals for the REML analysis will reduce the uncertainty in the heritability as compared to running the REML on a subset of individuals. I am guessing that subset heritabilities should be computable using the individual BLUPs, but I am unsure how to proceed.
Do you know if, and how, the individual BLUPs can be used to do this? Or if there is another method that would work well?
Thank you very much, Carolyn Jumper
|
|
|
Post by Zhihong Zhu on Jun 1, 2014 6:03:51 GMT
Please try "--keep" to estimate the heritability of the subset, i.e. "gcta --grm-bin grm_file --keep subset_file --reml --out output_file"
|
|
|
Post by carolynjumper on Jun 1, 2014 21:02:04 GMT
Thank you for the reply, but --keep does not do what I want. Using --keep in the GRM creation phase estimates the random effects using data only for the subset of individuals. I want to estimate the random effects on the whole dataset, but calculate heritability for a subset.
|
|