varCompCI: Variance Component Confidence Intervals

Description

varCompCI provides confidence intervals for the variance component estimates found using fitNullMM. The confidence intervals can be found on either the original scale or for the proportion of total variability explained.

Usage

varCompCI(nullMMobj, prop = TRUE)

Arguments

nullMMobj

A null model object returned by fitNullMM.

prop

A logical indicator of whether the point estimates and confidence intervals should be returned as the proportion of total variability explained (TRUE) or on the orginal scale (FALSE).

Value

prints a table of point estimates and 95% confidence interval limits for each estimated variance component.

Details

varCompCI takes the object returned by fitNullMM as its input and returns point estimates and confidence intervals for each of the random effects variance component estimates. If a kinship matrix or genetic relationship matrix (GRM) was included as a random effect in the model fit using fitNullMM, then this function can be used to provide a heritability estimate when prop is TRUE.

Examples

Run this code

# file path to GDS file
gdsfile <- system.file("extdata", "HapMap_ASW_MXL_geno.gds", package="GENESIS")
# read in GDS data
HapMap_geno <- GdsGenotypeReader(filename = gdsfile)
# create a GenotypeData class object
HapMap_genoData <- GenotypeData(HapMap_geno)
# load saved matrix of KING-robust estimates
data("HapMap_ASW_MXL_KINGmat")

# run PC-AiR
mypcair <- pcair(genoData = HapMap_genoData, kinMat = HapMap_ASW_MXL_KINGmat, 
                divMat = HapMap_ASW_MXL_KINGmat)
                
# run PC-Relate
mypcrel <- pcrelate(genoData = HapMap_genoData, pcMat = mypcair$vectors[,1],
        		training.set = mypcair$unrels)
close(HapMap_genoData)

# generate a phenotype
set.seed(4)
pheno <- 0.2*mypcair$vectors[,1] + rnorm(mypcair$nsamp, mean = 0, sd = 1)

# make ScanAnnotationDataFrame
scanAnnot <- ScanAnnotationDataFrame(data.frame(scanID = mypcrel$sample.id, 
              pc1 = mypcair$vectors[,1], pheno = pheno))

# make covMatList
covMatList <- list("Kin" = pcrelateMakeGRM(mypcrel))

# fit the null mixed model
nullmod <- fitNullMM(scanData = scanAnnot, outcome = "pheno", covars = "pc1", covMatList = covMatList)

# find the variance component CIs
varCompCI(nullmod, prop = TRUE)
varCompCI(nullmod, prop = FALSE)