Introduction

Code Dependencies

Usage

Heritability estimation of Dichotomous Phenotypes (LTMH)

# LTMH for random sample
 LTMH(model,data,init_beta=NULL,init_h2,V,famid,prev,max.iter=100,max.sub.iter=50,n.cores=1)
# LTMH for ascertained sample
 LTMH.asc(model,data,init_beta=NULL,init_h2,V,famid,prev,max.iter=100,n.cores=1,proband)
  • model : An object of class formula which is a symbolic description of the model to be fitted.
  • data : A data frame containing the variables in the model. Dependent variable in the model should be coded as 0 and 1 for unaffected and affected subjects, repelctively.
  • init_beta : Initial values for beta. It should be a numeric vector whose dimension is equal to that of beta. If no values or NULL were provided, coefficient estimates of a probit model obtained from glm.
  • init_h2 : Initial values for heritability on liability scale. It should be a numeric.
  • V : Generic relationship matrix such as kinship coefficient matrix. It should be a matrix which elements are numeric.
  • famid : Family ID for subjects. It should be a vector.
  • prev : Prevalence of a disease. It should be a numeric.
  • max.iter : Maximum number of iterations for EM (Expectation-maximization) algorithm. Default is 100.
  • max.sub.iter : Maximum number of sub-iterations for NR (Newton-Rahpson) algorithm. Default is 50.
  • n.cores : Number of cores for parallel computing. Unfortunatedly, it is not supported on Windows.
  • proband : A vector containing proband information which is coded as 0 and 1 for non-proband and proband, respectively.

Conditional Expected Score Test (CEST)

# CEST for heritability
 CEST.h2(model,data,init_beta=NULL,V,famid,prev,n.cores=1,proband=NULL)
# CEST for beta
 CEST.beta(model,data,test.beta,init_h2,V,famid,prev,max.iter=100,n.cores=1,proband=NULL)
  • model : An object of class formula which is a symbolic description of the model to be fitted.
  • data : A data frame containing the variables in the model. Dependent variable in the model should be coded as 0 and 1 for unaffected and affected subjects, repelctively.
  • test.beta : A vector of variable names to be tested.
  • init_beta : Initial values for beta. It should be a numeric vector whose dimension is equal to that of beta. If no values or NULL were provided, coefficient estimates of a probit model obtained from glm.
  • init_h2 : Initial values for heritability on liability scale. It should be a numeric.
  • V : Generic relationship matrix such as kinship coefficient matrix. It should be a matrix which elements are numeric.
  • famid : Family ID for subjects. It should be a vector.
  • prev : Prevalence of a disease. It should be a numeric which is ranged from 0 to 1.
  • max.iter : Maximum number of iterations for EM (Expectation-maximization) algorithm. Default is 100.
  • n.cores : Number of cores for parallel computing. Unfortunatedly, it is not supported on Windows.
  • proband : A vector containing proband information which is coded as 0 and 1 for non-proband and proband, respectively. If families were randomly selected, input NULL.

Generating the Simulation Data

# Generating nuclear families with a one main genetic effect (SNP)
 genNucFam(totalfam,num.fammem=c(3,4,5,6),prob.fammem=c(0.2,0.3,0.3,0.2),MAF,h2,ha2,prev,n.cores=1)
  • totalfam : Number of total families.
  • num.fammem : A vector of the number of family members to be generated. Default is c(3,4,5,6).
  • prob.fammem : A vector of the corresponding probabilities of num.fammem. Each element should be ranged in from 0 to 1 and its summation should be equal to 1. Default is c(0.2,0.3,0.3,0.2).
  • MAF : Minor allele frequency for a SNP.
  • h2 : Heritability which is ranged from 0 to 1.
  • ha2 : A relative proportion of variance explained by the disease SNP. It should be ranged from 0 to 1.
  • prev : Prevalence of a disease. It should be a numeric which is ranged from 0 to 1.
  • n.cores : Number of cores for parallel computing. Unfortunatedly, it is not supported on Windows.

Output

LTMH and LTMH.asc return an object of class list containing the following compoenents:

CEST.h2 and CEST.beta return an object of class data.frame containing the following columns:

Examples

Random families

  • 500 families were randomly generated with heritability of 0.2 and prevalence of 0.1.
  • Random_sp.txt : Datasets for 500 families which are randomly selected.
  • Random_kinship.txt : Kinship coefficients matrix for 500 families which are randomly selected.
## We recommend to run following example on device which can use multiple CPUs. It takes much time.
source("LTMH_Sourcecode.R")
dataset <- read.table("Random_sp.txt",head=T,stringsAsFactor=F)
V <- as.matrix(read.table("Random_kinship.txt",head=F))

LTMH(model=Y~snp-1,
     data=dataset,
     init_h2=0.2,
     V=V,
     famid=dataset$FID,
     prev=0.1)

CEST.h2(model=Y~snp-1,
        data=dataset,
        init_beta=NULL,
        V=V,
        famid=dataset$FID,
        prev=0.1)

Ascertained families

  • 500 families with affected proband were generated with heritability of 0.4 and prevalence of 0.1.
  • Ascertained_sp.txt : Datasets for 500 families with affected proband.
  • Random_kinship.txt : Kinship coefficients matrix for 500 families with affected proband.
## We recommend to run following example on device which can use multiple CPUs. It takes much time.
source("LTMH_Sourcecode.R")
dataset <- read.table("Ascertained_sp.txt",head=T,stringsAsFactor=F)
V <- as.matrix(read.table("Random_kinship.txt",head=F))

LTMH.asc(model=Y~snp-1,
     data=dataset,
     init_h2=0.4,
     V=V,
     famid=dataset$FID,
     prev=0.1,
     proband=dataset$ind)

CEST.h2(model=Y~snp-1,
        data=dataset,
        init_beta=NULL,
        V=V,
        famid=dataset$FID,
        prev=0.1,
        proband=dataset$ind)