iQRAT: Integrated Quantile Rank Test for group-wise joint effect of...

In tianyingw/iQRAT: Integrated Quantile RAnk Test

Integrated Quantile Rank Test for group-wise joint effect of rare and common variants in sequencing study

Description

This function implements the efficient quantile rank group-wise test, especially designed for testing the joint effect of rare and common variants in sequencing study. Comparing to existing methods, it tests integrated associations across the entire response distribution, rather than only focusing on mean or variability. This method does not have assumptions for error distributions. Thus, no quantile and rank normalization is required for data pre-processing. Note that when applying to univariate, no weights will be assigned.

Usage

iQRAT(
  X,
  C,
  v,
  cutoff = NULL,
  weight = TRUE,
  method.type = "S",
  w = NULL,
  SingleScore = NULL,
  RareOnly = FALSE,
  w.dbeta = NULL,
  acc = 1e-09
)

Arguments

X	A n by p matrix of genotype.
C	A vector or matrix of covariates.
v	Null model fitted based on quantile process.
cutoff	A number for separating common and rare variants. If not assigned, the default cutoff is 1/√(2*n).
weight	A logical variable indicating if different weights will be assigned to common and rare variants. The default is TRUE.
method.type	A character, either "S" for Q_S or "B" for Q_B.
w	A vector with length p. It could be functional annotation or other user-specified weights for variants. If w is NULL, the default beta density weight will apply.
SingleScore	Either "Wilcoxon", "Normal", "Lehmann" or "InverseLehmann" if you want to use only one weight function; otherwise, "NULL" represents a combination of four weight functions will be used.
RareOnly	A logical variable indicating if only consider rare variants effect. The default is FALSE. Once RareOnly is set to be TRUE, the cutoff value will be ignored.
w.dbeta	Two parameters for the beta density. The default value is NULL.

Value

A list contains p value, case and method type. Case indicates whether X is common, rare or mix.

References

Wang, T., Ionita-Laza, I. and Wei, Y. Integrated Quantile RAnk Test (iQRAT) for gene-level associations. Seunggeun Lee and with contributions from Larisa Miropolsky and Michael Wu.(2017) SKAT: SNP-Set (Sequence) Kernel Association Test.

Examples

rm(list=ls())
library(quantreg)
library(SKAT)
data("SampleData")

# Step 1: fit null model
null.fit = Null_model(Y = SampleData$y, C = SampleData$c)
# Step 2: run the test, p value will return

# SKAT version iQRAT
test.iQRAT1 = iQRAT(X = SampleData$x, C = SampleData$c, v =  null.fit, method.type = "S")

# If you want to specify weights
w = dbeta(colMeans(SampleData$x)/2,0.5,0.5) # we use beta density as an example
test.iQRAT2 = iQRAT(X = SampleData$x, C = SampleData$c, v =  null.fit, method.type = "S", w = w)

# Burden version iQRAT
test.iQRAT3 = iQRAT(X = SampleData$x, C = SampleData$c, v =  null.fit, method.type = "B")

tianyingw/iQRAT documentation built on Aug. 22, 2022, 11:25 a.m.