katya
New Member
Posts: 1
|
Post by katya on Jun 4, 2015 19:00:48 GMT
Dear all, I am new to using GCTA and obtained a confusing result when trying to partition heritability by chr. The overall heritability is h2=0.31, se=.17, p=0.029. When I run REML for each chromsome separately, heritability for each chr equals to h2=0.31, se=.17, p=0.029: Chr V(G)/Vp SE Pval
1 0.310149 0.167439 0.03035
2 0.310149 0.167439 0.03035
3 0.310149 0.167439 0.03035
... 22 0.310149 0.167439 0.03035
23 0.310149 0.167439 0.03035
However, when I run a join REML, all chromosomes except for chr2, have the following values h2=0.014, se=.000437, which would all sum up to 0.31: V(G1)/Vp 0.014098 0.000437
V(G2)/Vp 0.000001 0.168501
V(G3)/Vp 0.014098 0.000437
V(G4)/Vp 0.014098 0.000437
V(G5)/Vp 0.014098 0.000437 ...
What may be the issue here?
|
|
|
Post by Zhihong Zhu on Jun 5, 2015 8:20:30 GMT
Hi there,
I think there would be colinearity among your GRMs. Multiple publications indicate that SNP heritability is proportional to length of chromosome.
I'm not sure how you estimated GRM for each chromosome. But one approach is
gcta64 --bfile geno --chr 1 --make-grm --out chr1 gcta64 --bfile geno --chr 2 --make-grm --out chr2 ...
Cheers, Zhihong
|
|