runTypeI.
Random phenotypes are used to survey p-values under the null hypothesis (SNPs are not associated phenotype),
and genome-wide significance thresholds for single-SNP approach and GCDH are calculated by a user given
alpha-level.
A custom phe_fun is supplied
for simulating a phenotype associated with a certain pair of SNPs.
Total number of such simulations is set by the n_simu parameter. In each simulation 4 p-values are generated:
gcdhPower(rbed_info, n_shift, n_simu, maf_min, maf_max, r_limit, beta, collapse_matrix = NULL, dist_threshold = 5e+05, alpha_level = 0.05)n_shift for each GCDH run.P_GCDH: p-values from GCDH.
P_(single,no causal): p-values from single-SNP approach when causal SNPs are untyped.
P_(GCDH,no causal): p-values from GCDH when causal SNPs are untyped.
When all simulations are finished, 4 vectors of p-values are obtained: P_single_vec, P_GCDH_vec, P_(single,no causal)_vec, P_(GCDH,no causal)_vec. The power for each of the category (single-SNP, single-SNP without causal genotypes, GCDH, GCDH without causal genotypes) are proportions of these vectors that are more significant than the genome-wide significance thresholds we have obtained.