AM.SIMULATE: AM.SIMULATE

In yoki5348/VT_SEM: Vertical-Transmission Structural Equation Modeling (VT-SEM)

Description

Simulate trio dataset for the analysis of VT-SEM

Arguments

CV.INFO

Information of causal variants, must be 2 columned data frame. Column 1: Minor allele frequencies (MAF) of causal variants, Column 2: Effect sizes of causal variants

H2.TO	Initial heritability
NUM.GENERATIONS	Number of generations
POP.SIZE	Population sizes
MATE.COR	Assortative mating correlation : If there is no AM, MATE.COR=0
AVOID.INB	Bullean which indicates whether avoiding inbreeding or not (default if FALSE)
SAVE.EACH.GEN	Bullean which indicates whether saving whole dataset from initial generation to current generation or only current generation.
SAVE.COVS	Bulean which indicates saving covariances of random variables or not.
SEED	Set seed value.
VF.T0,PROP.H2.LATENT	Initial variance induced by vertical transmission
Unequal_AM	Bulean which indicates using disequilibrium AM assumption or not.

Value

SUMMARY.RES	Summary the simulation results from first generation to last.
XO	Generated genotype dataset of observed polygenic genetic scores (Additive coding)
XL	Generated genotype dataset of latent polygenic genetic scores (Additive coding)
PHEN	Generated phenotype dataset
HISTORY	Generated dataset which includes whole generation
COVARIANCES	Covariance matrices of phenotype dataset for each generation

Author(s)

Yongkang Kim [e-mail], Jared Balbona, and Matthew C Keller

References

Tahmasbi, Rasool, and Matthew C. Keller. "GeneEvolve: a fast and memory efficient forward-time simulator of realistic whole-genome sequence and SNP data." Bioinformatics 33.2 (2017): 294-296.

Examples

num.cvs <- 100 #The number of causal variants
am <-  0.25 # assortative mating parameter, mu
herit0 <- 0.5 #Inital heritability
vf0 <- 0.15 #Initial variance of vertical transmitted environment factor
######ve0=1-vf0-herit0, so herit0+vf0 must be less than 1
Unequal_AM=TRUE # TRUE: Generating disequilibrium AM dataset, FALSE: Generating equilibrium AM dataset
prop.h2.latent <- 0.5 #Proportion of heritability

seed <- 1234567
max.cores <- 1 ##Number of used cores for OpenMx
num.gen <- 20 #Number of generating generation
pop.size <- 1000 #The number of independent trio dataset
num.its <- 1000
RUN.MARKERS <- FALSE  #whether to only consider GRMs built from CVs (FALSE) or both CVs and SNPs (TRUE)
avoid.inb <- FALSE
save.covariances <- TRUE
save.history <- TRUE

#AM Simulation wildcards
MIN.MAF <- .1
MAX.MAF <- .50

#CVs and their effect sizes
MAF.VECTOR <- runif(num.cvs,MIN.MAF,MAX.MAF)  #Can change the distribution of MAFs here
GENTP.VAR <- MAF.VECTOR*(1-MAF.VECTOR)*2
#ALPHA.VECTOR <- rnorm(CVS,0,sqrt(1/(CVS*GENTP.VAR))) #Can change the distribution of effect sizes here - random effects
ALPHA.VECTOR <- sample(c(-1,1),num.cvs,replace=TRUE)*sqrt(1/(num.cvs*GENTP.VAR)) #Can change the distribution of effect sizes here - fixed f'n of MAF
CV.INFO <- data.frame(MAF=MAF.VECTOR,alpha=ALPHA.VECTOR) #we'll use this for both the observed and latent

AM.DATA <- AM.SIMULATE(CV.INFO=CV.INFO, H2.T0=herit0, NUM.GENERATIONS=num.gen, POP.SIZE=pop.size*3, MATE.COR=am, AVOID.INB=avoid.inb, SAVE.EACH.GEN=save.history, SAVE.COVS=save.covariances, SEED=seed, VF.T0=vf0,PROP.H2.LATENT=prop.h2.latent,Unequal_AM=Unequal_AM) #Dataset generation

yoki5348/VT_SEM documentation built on July 24, 2021, 5:10 p.m.