Simulate multivariate GWAS Data with Specified Direct Effects

Usage

sim_mv_determined(
  N,
  direct_SNP_effects_joint,
  geno_scale,
  pheno_sd,
  G = 0,
  est_s = FALSE,
  R_obs = NULL,
  R_E = NULL,
  R_LD = NULL,
  af = NULL
)

Arguments

N: GWAS sample size. N can be a scalar, vector, or matrix. If N is a scalar, all GWAS have the same sample size and there is no overlap between studies. If N is a vector, each element of N specifies the sample size of the corresponding GWAS and there is no overlap between studies. If N is a matrix, N_ii specifies the sample size of study i and N_ij specifies the number of samples present in both study i and study j. The elements of N must be positive but non-integer values will not generate an error.
direct_SNP_effects_joint: Matrix of direct variant effects. Should be variants by traits.
geno_scale: Genotype scale of provided effects. Either "allele" or "sd".
pheno_sd: Phenotype standard deviation, a scalar or vector of length number of traits.
G: Matrix of direct effects. Rows correspond to the 'from' trait and columns correspond to the 'to' trait, so G[1,2] is the direct effect of trait 1 on trait 2. G should have 0 on the diagonal. Be sure that G is on the same scale as the effect sizes.
est_s: If TRUE, return estimates of se(`beta_hat`). If FALSE, the exact standard error of `beta_hat` is returned. Defaults to FALSE.
R_obs: Total observational correlation between traits. R_obs won't impact summary statistics unless there is sample overlap. See Details for default behavior.
R_E: Total correlation of the environmental components only. R_E and R_obs are alternative methods of specifying trait correlation. Use only one of these two options. R_E may be phased out in the future.
R_LD: Optional list of LD blocks. R_LD should have class list. Each element of R_LD can be either a) a matrix, b) a sparse matrix (class dsCMatrix) or c) an eigen decomposition (class eigen). All elements should be correlation matrices, meaning that they have 1 on the diagonal and are positive definite. See Details and vignettes.
af: Optional vector of allele frequencies. If R_LD is not supplied, af can be a scalar, vector or function. If af is a function it should take a single argument (n) and return a vector of n allele frequencies (See Examples). If R_LD is supplied, af must be a vector with length equal to the size of the supplied LD pattern (See Examples).

Value

A sim_mv function. See ?sim_mv for details.

Details

A wrapper for sim_mv. See ?sim_mv and the "Providing an Exact Set of Direct Effects" section of the Effect Size vignette.

Examples

G <- matrix(c(0, 0.5, 0, 0), nrow = 2, byrow =T)
my_effects <- matrix(0, nrow = 10, ncol = 2)
my_effects[c(1, 5),1] <- c(-0.008, 0.01)
my_effects[c(3, 6, 9), 2] <- c(-0.02, 0.06, 0.009)
my_effects
#>         [,1]   [,2]
#>  [1,] -0.008  0.000
#>  [2,]  0.000  0.000
#>  [3,]  0.000 -0.020
#>  [4,]  0.000  0.000
#>  [5,]  0.010  0.000
#>  [6,]  0.000  0.060
#>  [7,]  0.000  0.000
#>  [8,]  0.000  0.000
#>  [9,]  0.000  0.009
#> [10,]  0.000  0.000
# for fun, lets include some sample overlap
N <- matrix(c(40000, 10000, 10000, 20000), nrow = 2)
sim_dat <- sim_mv_determined(N = N,
                              direct_SNP_effects_joint = my_effects,
                              geno_scale = "sd",
                              pheno_sd = 1,
                              G=G,
                              est_s = TRUE)
#> SNP effects provided for 10 SNPs and 2 traits.

sim_dat$direct_SNP_effects_joint
#>         [,1]   [,2]
#>  [1,] -0.008  0.000
#>  [2,]  0.000  0.000
#>  [3,]  0.000 -0.020
#>  [4,]  0.000  0.000
#>  [5,]  0.010  0.000
#>  [6,]  0.000  0.060
#>  [7,]  0.000  0.000
#>  [8,]  0.000  0.000
#>  [9,]  0.000  0.009
#> [10,]  0.000  0.000
sim_dat$beta_joint
#>         [,1]   [,2]
#>  [1,] -0.008 -0.004
#>  [2,]  0.000  0.000
#>  [3,]  0.000 -0.020
#>  [4,]  0.000  0.000
#>  [5,]  0.010  0.005
#>  [6,]  0.000  0.060
#>  [7,]  0.000  0.000
#>  [8,]  0.000  0.000
#>  [9,]  0.000  0.009
#> [10,]  0.000  0.000
sim_dat$Sigma_G
#>          [,1]     [,2]
#> [1,] 0.000164 0.000082
#> [2,] 0.000082 0.004122