Define hyper parameters — define_hyper

Define hyper parameters for simulating mendelian randomization (MR) data.

Usage

define_hyper_parameters(d, s, p, r2_g_x, r2_u_x, r2_u_y, beta_x_y)

d: (Dimension) Number of G's. It should be a positive integer
s: (Sparsity) Proportion of zero effect G's. It should be between 0 and 1
p: (Skewness) Minor allele frequency of G'. It should be between 0 and 0.5
r2_g_x: (Instrument strength) Variance in X explained per non-zero effect G. It should be between 0 and 1 / ceiling((1 - s) * d)
r2_u_x: (Confouding level) Variance in X explained by U. It should be between 0 and 1
r2_u_y: (Confouding level) Variance in Y explained by U. It should be between 0 and 1
beta_x_y: (Target causal effect) Causal effect of X on Y. It should be between - 1 / sqrt(r2_g_x) and 1 / sqrt(r2_g_x). Since all variables are standardized, a more reasonable range is between -1 and 1

An object of class hyper_parameters

Additional constraints may be posed to the hyper parameter values in order to guarantee that the simulated data is valid.