reconstruct joint model ioutput for single snp. only vb1b2 really needed
qq.onesnp(b1, b2, vb1, vb2, fx, v1, v2, eta, N)
| b1 | coef for snp X, trait 1 |
|---|---|
| b2 | coef for snp X, trait 2 |
| vb1 | var(coef) for snp X, trait 1 |
| vb2 | var(coef) for snp X, trait 2 |
| fx | MAF of X |
| v1 | variance of trait 1 |
| v2 | variance of trait 2 |
| eta | cor(trait1, trait2) |
| N | number of samples |
var(b1,b2)
The SNP is X, the traits are 1 and 2
note in this case, the coefficients, and their variances, are equal in single or joint models. it is just the var(b1,b2) term we need.
This provides cor(b1,b2), and must be multiplied by sqrt(v(b1)*v(b2)) to get var(b1,b2)