GWAS for Studies having Repeated Measurements on Related Individuals

Share:

Description

It is used to perform genome-wide association studies on individuals that are both related and have repeated measurements. The function computes score statistic based p-values for a linear mixed model including random polygenic effects and a random effect for repeated measurements. A p-value is computed for each marker and the null hypothesis tested is a zero additive marker effect.

Usage

1
2
rGLS(formula.FixedEffects = y ~ 1, genabel.data, phenotype.data,
  id.name = "id", GRM = NULL, V = NULL, memory = 1e+08)

Arguments

formula.FixedEffects

Formula including the response variable and cofactors as fixed effects.

genabel.data

An GenABEL object including marker information. This object has one observtion per individuals.

phenotype.data

A data frame including the repeated observations and IDs.

id.name

The column name of the IDs in phen.data

GRM

An optional genetic relationship matrix (GRM) can be included as input. Otherwise the GRM is computed within the function.

V

An optional (co)variance matrix can be included as input. Otherwise it is computed using the hglm function.

memory

Used to optimize computations. The maximum number of elements in a matrix that can be stored efficiently.

Details

A generalized squares (GLS) is fitted for each marker given a (co)variance matrix V. The computations are made fast by transforming the GLS to an ordinary least-squares (OLS) problem using an eigen-decomposition of V. The OLS are computed using QR-factorization. If V is not specified then a model including random polygenic effects and permanent environmental effects is fitted (using the hglm package) to compute V. A GenABEL object (scan.gwaa class) is returned (including also the hglm results). Let e.g. GWAS1 be an object returned by the rGLS function. Then a Manhattan plot can be produced by calling plot(GWAS1) and the top SNPs using summary(GWAS1). Both of these functions are generic GenABEL functions.
The results from the fitted linear mixed model without any SNP effect included are produced by calling summary(GWAS1@call$hglm).

Author(s)

Lars Ronnegard

Examples

1
2
3
4
5
6
7
data(Phen.Data) #Phenotype data with repeated observations
 data(gen.data) #GenABEL object including IDs and marker genotypes
 GWAS1 <- rGLS(y ~ age + sex, genabel.data = gen.data, phenotype.data = Phen.Data)
 plot(GWAS1, main="")
 summary(GWAS1)
 #Summary for variance component estimation without SNP effects
 summary(GWAS1@call$hglm)