setup for computing the loglikelihood for the model
Assumes equally spaced covariates
Y \sim N(X_k\beta + X \gamma, \sigma_Y^2 I_n \times n)
\gamma \sim N(u, \tau^2 I_p \times p)
u = [u_1,...,u_n_c]
u_i \sim N(\mu_0, \sigma^2_u 1
@param Y (n x 1) the data
@param X (n x p) the gene matrixs
@param Xk (n x k) the covariates (for \beta) if no covariates set NULL
@param nc (int - matrix) number of chromosones, if null simpler model