generate_dependence_from_annots: Generate epigenetic marks, SNPs and phenotypes with...

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

Generate epigenetic marks, SNPs and phenotypes with epigenome-driven genetic associations

Description

Generate epigenetic marks, SNPs and phenotypes with epigenome-driven genetic associations

Usage

generate_dependence_from_annots(
  n,
  n_loci,
  mean_locus_size,
  p0,
  rho_min_x,
  rho_max_x,
  n_modules,
  mean_module_size,
  rho_min_y,
  rho_max_y,
  r,
  r0,
  prop_act,
  max_tot_pve,
  annots_vs_indep = 1,
  min_dist = 0,
  maf_thres = 0.05,
  max_nb_act_snps_per_locus = 3,
  vec_q = NULL,
  real_snp_mat = NULL,
  real_annot_mat = NULL,
  sd_act_beta = NULL,
  q_pres_annot_loci = NULL,
  bin_annot_freq = 0.05,
  candidate_modules_annots = NULL,
  tpois_lam_act_annots_mm = 1,
  sd_act_prob = 1,
  sd_pat = 1,
  sd_err = 1,
  rbeta_sh1_rr = 1,
  n_cpus = 1,
  maxit = 10000,
  module_specific = FALSE,
  user_seed = NULL,
  return_patterns = FALSE
)

Arguments

n	Number of observations.
n_loci	Number of loci.
mean_locus_size	Mean locus size (drawn from a Poisson distribution).
p0	Minimum number of active SNPs (i.e., associated with at least one phenotype).
rho_min_x	Minimum autocorrelation value for blocks of SNPs in linkage-disequilibrium.
rho_max_x	Maximum autocorrelation value for blocks of SNPs in linkage-disequilibrium.
n_modules	Number of modules of phenotypes.
mean_module_size	Mean module size (drawn from a Poisson distribution).
rho_min_y	Minimum equicorrelation value for the phenotypes in a given module. If NULL, independent SNPs simulated.
rho_max_y	Minimum equicorrelation value for the phenotypes in a given module. If NULL, independent SNPs simulated.
r	Total number of epigenetic annotations.
r0	Number of epigenetic annotations which trigger genetic associations.
prop_act	Approximate proportion of associated SNP-phenotype pairs.
max_tot_pve	Maximum variance explained by the SNPs for a given phenotype.
annots_vs_indep	Proportion of active SNPs whose effects are triggered by epigenetic marks. Default is 1, for all effects triggered by the epigenome.
min_dist	Minimum distance between each pair of loci (in terms of number of SNPs). Default is 0 for no distance enforced.
maf_thres	Minor allele frequency threshold (applied for both supplied and simulated SNPs). Default is 0.05.
max_nb_act_snps_per_locus	Maximum number of active SNPs per locus. Default is 3.
vec_q	Exact module sizes. Either mean_module_size or vec_q must be NULL. Default is NULL.
real_snp_mat	Matrix of real SNPs supplied by the user. Default is NULL for simulated SNPs under the Hardy-Weinberg assumption.
real_annot_mat	Matrix of real epigenetic annotations supplied by the user. Default is NULL for simulated binary annotations.
sd_act_beta	Standard deviation of the simulated QTL effects. Either sd_act_beta or max_tot_pve must be NULL. Default is NULL.
q_pres_annot_loci	Quantile for selecting annotations which concern most loci (i.e., at least one SNP in each locus). Should be large so enough candidate active SNPs are available when annots_vs_indep is large. Default is NULL.
bin_annot_freq	Minimum frequency of SNPs concerned by a given annotation. Default is 0.05.
candidate_modules_annots	The subset of module ids where all associations are triggered by annotations. The complement are the modules where associations are independent of the annotations. Default is NULL for all modules used as active modules. If n_modules is large, specify a smaller subset of modules, as the mapping may fail otherwise.
tpois_lam_act_annots_mm	Zero-truncated Poisson parameter for drawing the number of active annots per module. Default is 1.
sd_act_prob	Standard deviation for the effects of SNPs and annotations. Default is 1.
sd_pat	Standard deviation for the randomness of the SNP-trait pattern. Default is 1.
sd_err	Response error standard deviation. Default is 1.
rbeta_sh1_rr	Beta distribution shape2 parameter for the proportion of responses associated with an active SNP (in a given module) rbeta_sh2_rr = 1 (default), so right skewed if rbeta_sh1_rr > 1.
n_cpus	number of CPUs to be used. Default is 1.
maxit	Maximum number of iterations for the repeat loops. Default is 1e4.
module_specific	Boolean specifying whether the epigenome activation is module-specific or not. Default is FALSE
user_seed	Seed set for reproducibility. Default is NULL, no seed set.
return_patterns	Boolean specifying whether the simulated SNP-phenotype association pattern and active annotation variables.

Value

A list containing matrices of

snps	Matrix containing the simulated or supplied SNP data.
annots	Matrix containing the simulated or supplied epiegenetic annotation data.
phenos	Matrix containing the simulated phenotypic data.
pat	If return_patterns is TRUE, simulated SNP-phenotype association pattern.
beta	If return_patterns is TRUE, simulated SNP-phenotype regression coefficients.
active_annots	If return_patterns is TRUE, active annotation variables.

Examples

user_seed <- 123; set.seed(user_seed)

# Number of samples
#
n <- 500

# Loci
#
n_loci <- 20
mean_locus_size <- 100
p0 <- 10

# Modules of traits
#
n_modules <- 5
mean_module_size <- 50

# Autocorrelation within loci and equicorrelation within trait modules
#
rho_min_x <- rho_min_y <- 0.5
rho_max_x <- rho_max_y <- 0.9

# Annotations
#
r <- 200
r0 <- 10

# Association pattern
#
prop_act <- 0.1
max_tot_pve <- 0.5

list_assoc <- generate_dependence_from_annots(n, n_loci, mean_locus_size, p0,
                                              rho_min_x, rho_max_x,
                                              n_modules, mean_module_size,
                                              rho_min_y, rho_max_y, r, r0,
                                              prop_act, max_tot_pve,
                                              user_seed = user_seed)

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