gemma2 is an implementation in R of the GEMMA v 0.97 EM algorithm that is part of the GEMMA algorithm for REML estimation of multivariate linear mixed effects models of the form:
[vec(Y) = X vec(B) + vec(G) + vec(E)]
where (E) is a n by 2 matrix of random effects that follows the matrix-variate normal distribution
[G MN(0, K, V_g)]
where (K) is a relatedness matrix and (V_g) is a 2 by 2 covariance matrix for the two traits of interest.
Additionally, the random errors matrix (E) follows the distribution:
[E MN(0, I_n, V_e)]
and (G) and (E) are independent.
To install gemma2, use the devtools R package from CRAN. If you haven’t installed devtools, please run this line of code:
Then, run this line of code to install gemma2:
X. Zhou & M. Stephens. Efficient multivariate linear mixed model algorithms for genome-wide association studies. Nature Methods volume 11, pages 407–409 (2014). https://www.nature.com/articles/nmeth.2848