KMgene (version 1.3)

MFKMO: Optimal KM for Quantitative Traits in Multivariate Family GWAS Data (calculate p-value)

Description

This function (MFKMO) is used to perform optimal KM analysis for quantitative traits in GWAS multivariate family data.

Usage

MFKMO(
  obj,
  genotypes,
  weights = NULL,
  acc = 1e-04,
  acc2 = 1e-04,
  r.all = c(0, 0.25, 0.5, 0.75, 1),
  append.write = NULL,
  eq.gen.effect = F
)

Arguments

obj

results saved from MFKMO_Null_Model.

genotypes

1st column: gene name; 2nd column: snp name; 3rd-end columns: A matrix of genotypes for each subject (class: data.frame). The order of 3rd-end columns should match unique(yid). Coded as 0, 1, 2 and no missing. This genotype file can be a big file containing all genes or it can be files containing one single gene.

weights

1st column: gene name; 2nd column: snp name; 3rd column: A vector with the length equal to the number of variants in the test (class: data.frame). Default is Null indicating equal weight for all markers

acc

Accuracy of numerical integration used in Davies' method for individual r.all p-values. Default 1e-4.

acc2

Accuracy of numerical integration used in Davies' method for the final p-value. Default 1e-4.

r.all

A list of predefined proportion of linear kernel and burden test. When r.all=0, regular kernel machine test (MFKM); when r.all=1, burden test.

append.write

The name of pvalue output file. Write out p-values in real time. Don't need to wait until all genes are processed to get the final output.

eq.gen.effect

Whether assume equal genetic effects on different traits (default = False).

Value

output: optimal multivariate family KM (MF-KMO) p-value

Examples

# NOT RUN {
############################################################################
### Examples for Multivariate (two) Continuous Traits in Familial GWAS Data
############################################################################
######################
# using optimal KM ###
######################
### Subject IDs are numeric ###
data("MFKM_numID")
obj1 <- MFKMO_Null_Model(phenotype=mfkm_n_y$y, trait=mfkm_n_y$trait, yid=mfkm_n_y$id,
gid=mfkm_n_geneid$gid, fa=mfkm_n_geneid$fa, mo=mfkm_n_geneid$mo, covariates=NULL,
Ninitial=1)
pvalue1 <- MFKMO(obj=obj1, genotypes=mfkm_n_gene, weights=NULL)
# Read in a list of genes files instead of a big file containing all genes
obj <- MFKMO_Null_Model(phenotype=mfkm_n_y$y, trait=mfkm_n_y$trait, yid=mfkm_n_y$id,
gid=mfkm_n_geneid$gid, fa=mfkm_n_geneid$fa, mo=mfkm_n_geneid$mo, covariates=NULL,
Ninitial=1)
gene <- split(mfkm_n_gene, mfkm_n_gene[,1])
for (k in 1:2) {
  gene[[k]]$gene <- as.character(gene[[k]]$gene)
  pvalue1 <- MFKMO(obj=obj, genotypes=gene[[k]], weights=NULL)
}
### Subject IDs are character ###
data("MFKM_charID")
obj1 <- MFKMO_Null_Model(phenotype=mfkm_c_y$y, trait=mfkm_c_y$trait,
yid=as.character(mfkm_c_y$id),
gid=as.character(mfkm_c_geneid$gid), fa=as.character(mfkm_c_geneid$fa),
mo=as.character(mfkm_c_geneid$mo), covariates=NULL, Ninitial=1)
pvalue1 <- MFKMO(obj=obj1, genotypes=mfkm_c_gene, weights=NULL)
# }