Simulation of genotypes with family information (Fixed and non-fixed)

This function is used to simulate genotypes with accompanying phenotype information, as well as phenotype information on the case-control status of parents and possible siblings.

Usage

sim_genotypes_with_family(
  n,
  disease,
  path,
  n_sibs = NULL,
  overwrite = T,
  n_blocks = min(n, 500)
)

Arguments

n

Integer specifying the number of individuals/genotypes to simulate.

disease

A list with all the disease parameters.

path

Path to where .rds file should be saved, or where one is stored if overwriting existing .rds file (DO NOT SPECIFY FILE EXTENSION).

n_sibs

Integer value for how many sibling to produce for each genotype or vector containing values for how many sibs to sample from (ex. c(1,4,6) will produce genotypes randomly with 1, 4 or 6 siblings).

overwrite

Boolean value used to determine if existing .rds file with specified name should be overwritten (Default value TRUE).

n_blocks

Integer used to determine number of blocks to run simulation in (Default value is 500). Set higher if running into memory issues such as freezing or crashing. Setting n_blocks higher reduces the memory size of each block, but slightly slows the calculation time.

Value

Returns list object, refered to as an rds object, containing an FMB.code256 with genotypes, MAF tibble containing information on SNPs and FAM tibble containing phenotype information on genotypes as well as case-control status of parents and possible siblings.

Details

Simulating a 100.000x100.000 dataset will take up around 9.76 GB of space. Since the running time depends on a number of variables, such as the parallelization settings, core speed, core amount and sibling configuration, we cannot accurately give an estimation how long the simulation will take. Instead we simply warn the user that simulations might take upwards of multiple hours for large datasets such as a 100.000x100.000. The default n_blocks parameter has been set to 500 as this is the number at which a 100.000x100.000 with 2 siblings for each genotype will use a maximum of 2 GB of RAM for calculating a single block. Simulation can be performed using parallelization if a parallelization plan has been set prior to execution in the global environment. WARNING: using parallelization will, with a n_blocks of 500, use up to a maximum of 2 GB of RAM for EACH process when running a simulation of 100.000x100.000 with 2 siblings for each genotype.