Resample Summary Statistics for Existing Simulation Object

Usage

resample_sumstats(
  dat,
  N,
  R_LD = NULL,
  af = NULL,
  est_s = FALSE,
  geno_scale = NULL,
  new_env_var = NULL,
  new_h2 = NULL,
  new_R_E = NULL,
  new_R_obs = NULL
)

Arguments

dat: Object output by sim_mv
N: Sample size, scalar, vector, matrix. See ?sim_mv for more details.
R_LD: LD pattern (optional). See ?sim_mv for more details.
af: Allele frequencies. See ?sim_mv for more details.
est_s: Logical, should estimates of se(beta_hat) be produced.
geno_scale: Either "allele" or "sd". Specifies the scale of the effect sizes in the output data.
new_env_var: Optional. The environmental variance in the new population. If missing the function will assume the environmental variance is the same as in the old population.
new_h2: Optional. The heritability in the new population. Provide at most one of new_env_var and new_h2.
new_R_E: Optional, specify environmental correlation in the new population. If missing, the function will assume the environmental correlation is the same as in the original data.
new_R_obs: Optional, specify overall trait correlation in the new population. Specify at most one of new_R_E or new_R_obs. If missing, the function will assume the environmental correlation is the same as in the original data.

Details

This function can be used to generate new summary statistics for an existing simulation object. For a discussion of this function and resample_inddata, see the "Resampling" vignette.

Examples

# Use resample_sumstats to generate new GWAS results with the same effect sizes.
N <- matrix(1000, nrow = 2, ncol =2)
G <- matrix(0, nrow = 2, ncol = 2)
R_E <- matrix(c(1, 0.8, 0.8, 1), nrow = 2, ncol = 2)
# original data
dat <- sim_mv(N = N, J = 20000, h2 = c(0.4, 0.3), pi = 1000/20000,
               G = G, R_E = R_E)
#> SNP effects provided for 20000 SNPs and 2 traits.
# data for second GWAS
dat_new <- resample_sumstats(dat,
                             N = 40000)
#> I will assume that the environmental variance is the same in the old and new population.
#> I will assume that environmental correlation is the same in the old and new population. Note that this could result in different overall trait correlations.
#> Note that the phenotype in the new population has a different variance from the phenotype in the old population.
#> I will keep the phenotype on the same scale as the original data, so effect sizes in the old and new object are comparable. If you would like to rescale the phenotype to have variance 1, use rescale_sumstats.
#> Original data have effects on the per-genotype sd scale. I will assume that per-genotype sd effects are the same in the new and old populations.
#> SNP effects provided for 20000 SNPs and 2 traits.