Initialize a susie object using regression coefficients

susie_init_coef(coef_index, coef_value, p)

Arguments

coef_index: An L-vector containing the the indices of the nonzero coefficients.
coef_value: An L-vector containing initial coefficient estimates.
p: A scalar giving the number of variables.

Value

A list with elements alpha, mu and mu2 to be used by susie.

Examples

set.seed(1)
n = 1000
p = 1000
beta = rep(0,p)
beta[sample(1:1000,4)] = 1
X = matrix(rnorm(n*p),nrow = n,ncol = p)
X = scale(X,center = TRUE,scale = TRUE)
y = drop(X %*% beta + rnorm(n))

# Initialize susie to ground-truth coefficients.
s = susie_init_coef(which(beta != 0),beta[beta != 0],length(beta))
res = susie(X,y,L = 10,s_init=s)
#> Specified number of effects L = 10 is greater the number of effects 4 in input SuSiE model. The SuSiE model will be expanded to have 10 effects.