replicate_real_phenos: Generate phenotypes emulating real phenotypic data at hand.

In hruffieux/echoseq: Replication and Simulation of Genetic Variants, Molecular Expression Levels and Other Phenotypic Data

Generate phenotypes emulating real phenotypic data at hand.

Description

This function simulates phenotypes from real phenotypic data based on their sample correlation structure. If binary data are provided, will simulate latent Gaussian data from them (binary data can then be obtained using the generate_dependence function).

Usage

replicate_real_phenos(
  n,
  real_phenos,
  input_family = "gaussian",
  bl_lgth = NULL,
  d = NULL,
  n_cpus = 1,
  user_seed = NULL
)

Arguments

n	Number of observations.
real_phenos	Matrix of real phenotypes (rows observations, columns phenotypic variables), without missing values.
input_family	Phenotype distribution assumption for the phenotypes provided in real_phenos. Must be either "gaussian" or "binomial" for binary phenotypes.
bl_lgth	Number of variables per block for reproducing the dependence structure of real phenotypes. Must be between 2 and d. Must be small enough (e.g. 1000) for tractability reasons. Default is NULL for a single block.
d	Number of phenotypes. If NULL, the total number of phenotypes available in the real_phenos matrix is used (default).
n_cpus	Number of CPUs used when simulating phenotypes by blocks. Set to 1 for serial execution.
user_seed	Seed set for reproducibility. Default is NULL, no seed set.

Value

An object of class "sim_phenos".

phenos	Matrix containing the generated phenotypic data.
var_err	Vector containing the sample phenotypic variances.
ind_bl	List of length given by the number of blocks, containing the indices of the phenotypes in each block. Is NULL if cor_type is NULL.

See Also

convert_snps, generate_snps, replicate_real_snps, convert_phenos, generate_phenos, generate_dependence

Examples

user_seed <- 123; set.seed(user_seed)
n <- 500; d <- 1000
cor_type <- "equicorrelated"; vec_rho <- runif(100, min = 0.25, max = 0.95)

# Provided phenotypes assumed to be normally distributed
var_err <- runif(d, min = 0.1, max = 0.4)
list_fake_real_phenos <- generate_phenos(n, d, var_err, cor_type = cor_type,
                                         vec_rho = vec_rho, n_cpus = 1,
                                         user_seed = user_seed)
list_phenos <- replicate_real_phenos(n, list_fake_real_phenos$phenos,
                                     input_family = "gaussian", bl_lgth = 100,
                                     n_cpus = 1, user_seed = user_seed)

hruffieux/echoseq documentation built on Jan. 10, 2024, 10:06 p.m.