coxSKAT: SNP-set kernel association test for right-censored survival...

Description Usage Arguments Details Value Author(s) References Examples

Description

Tests for association between a set of common SNPS and a right-censored survival outcome. Warnings: (1) coxKM is meant for common genetic variants, (2) for very small p-values, it is necessary to increase no. of perturbations.

Usage

1
2
3
4
5
 coxKM(Z=NULL, U, Delta, X=NULL, gamma=NULL, kernel="linear", weights=NULL, 
	npert=10^4, npert.check=TRUE, npert.upper=10^8, npert.threshold=50,
	impute.method = "fixed", is_check_genotype=TRUE, 
	is_dosage=FALSE, missing_cutoff=0.15, SetID=NULL)
 

Arguments

X

is a nxR matrix of relevant covariates with each row as a different individual and each column as a separate covariate measurement. If no additional covariates are present, X can be left unspecified or left as NULL. Note that each column of X has to be a numerical variable, non-numerical variables have to be recoded appropriately before analysis. X should not include an intercept.

Z

is a nxS numeric genotype matrix with each row as a different individual and each column as a separate snp. Each genotype should be coded as 0, 1, 2, and 9 (or NA) for AA, Aa, aa, and missing, where A is a major allele and a is a minor allele. Missing genotypes will be imputed by the simple Hardy-Weinberg equilibrium (HWE) based imputation. If kernel matrix is supplied, Z is ignored and not used in testing.

U

is a nx1 vector containing the observed times. Note: U=min(C,T) where C = censoring time, T = survival time

Delta

is a nx1 vector containing the status/event indicator.

gamma

Unless X = NULL, gamma has to be supplied. gamma is the vector of coefficients from the null cox model corresponding to X. gamma <- coxph(Surv(U,Delta)~X)$coef

kernel

Type of kernel. kernel can be an nxn kernel matrix OR one of these six options: "linear.weighted", "linear", "IBS", "IBS.weighted", "quadratic" or "2wayIX". If an nxn kernel matrix is supplied, Z is ignored and is not used in testing.

weights

is a vector of length S of prespecified weights for the weighted kernels. Weights in coxKM are defined the same way as in SKAT. The kernel matrix of the weighted linear kernel is K=ZWWZ'.

npert

is the number of perturbations used to calculate p-value (default =10000), npert should be at least 1000. Note that how small the p-value can be is limited by the number of perturbations. If npert.check = FALSE, the smallest possible p-value is 0.5/npert. If npert.check = TRUE, the smallest possible p-value is 0.5/(ceiling(npert.upper/10^4)*10^4). For very small p-values, to obtain accurate p-values, it is necessary to increase the number of perturbations. See npert.check.

npert.check

TRUE/FALSE (default=TRUE). If npert.check=TRUE, coxKM first uses npert perturbations to obtain an initial p-value and checks to see if the initial p-value <= npert.threshold/npert. If the initial p-value <= npert.threshold/npert, then npert.upper perturbations is used to obtain a more accurate p-value. Setting npert.check=TRUE allows a larger number of perturbations to be used to obtain more accurate p-values only when it is necessary. For very small p-values, it may be necessary to further increase npert.upper.

npert.upper

default=10^8. Used only if npert.check=TRUE. See npert.check.

npert.threshold

default=50. Used only if npert.check=TRUE. See npert.check.

impute.method

a method to impute missing genotypes (default= "fixed"). "random" imputes missing genotypes by generating binomial(2,p) random variables (p is the MAF), and "fixed" imputes missing genotypes by assigning the mean genotype value (2p). If you use "random", you will have different p-values for different runs because imputed values are randomly assigned. Can use set.seed() to replicate results.

is_check_genotype

a logical value indicating whether to check the validity of the genotype matrix Z (default= TRUE). If you use non-SNP type data and want to run coxKM, please set it to FALSE. If you use SNP data or imputed data, please set it to TRUE. If is_check_genotype=FALSE, missing values in Z have to be coded only as NA since 9 will not be treated as a missing value.

is_dosage

a logical value indicating whether the matrix Z is a dosage matrix (default= FALSE). If is_dosage=TRUE, “is_check_genotype” and “impute.method” will be ignored and coxKM will check the genotype matrix and set impute.method="fixed". Note that coxKM will also treat 9 as missing in Z.

missing_cutoff

a cutoff of the missing rates of SNPs (default=0.15). Any SNPs with missing rates higher than cutoff will be excluded from the analysis.

SetID

SetID.

Details

If kernel is not a matrix and Z is supplied, and either is_check_genotype=TRUE OR is_dosage=TRUE, coxKM will check the Z matrix for missing values (missing values must be coded either as NA or 9) and apply imputation. If you are using coxKM for non-SNP/dosage data, set is_check_genotype=FALSE and is_dosage=FALSE, in which case missing values must be coded as NA (9 is not considered a missing value).

Value

p.value

the p-value of coxKM based on resampling. Note that if the p-value takes on the smallest possible value based on the number of perturbations, it may be necessary to increase npert and npert.upper. See npert.check.

Q

the unscaled score test statistic of coxKM.

n.marker.test

no. of SNPs used for testing, <=S.

n.indiv

n = no. of samples

df

the estimated degrees of freedom of the test statistic (for reference only, not used in association testing) .

Author(s)

Xinyi (Cindy) Lin, Qian Zhou

References

Lin X, Cai T, Wu M, Zhou Q, Liu G, Christiani D and Lin X. 2011. Survival Kernel Machine SNP-set Analysis for Genome-wide Association Studies. Genetic Epidemiology 35:620-31. doi: 10.1002/gepi.20610

Cai T, Tonini G and Lin X. 2011. Kernel machine approach to testing the significance of multiple genetic markers for risk prediction. Biometrics, 67:975-86. doi:10.1111/j.1541-0420.2010.01544.x

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
data(examplesnpset, examplecovariates, examplephenotype1, examplephenotype2, examplephenotype3)

Z <- as.matrix(examplesnpset)
X <- as.matrix(examplecovariates)
phenotype1 <- examplephenotype1
phenotype2 <- examplephenotype2
phenotype3 <- examplephenotype3

set.seed(1)

#---------------------------------------------------------------------
# coxKM without covariates
#---------------------------------------------------------------------
coxKM(Z=Z, U=phenotype1$time, Delta=phenotype1$event, kernel="IBS")
coxKM(Z=Z, U=phenotype1$time, Delta=phenotype1$event, kernel="linear")
coxKM(Z=Z, U=phenotype3$time, Delta=phenotype3$event, kernel="IBS")
coxKM(Z=Z, U=phenotype3$time, Delta=phenotype3$event, kernel="linear")


#---------------------------------------------------------------------
# coxKM with covariates
#---------------------------------------------------------------------
Gamma <- coxph(Surv(phenotype2$time, phenotype2$event)~X)$coef
Gamma
coxKM(Z=Z, U=phenotype2$time, Delta=phenotype2$event, X=X, gamma=Gamma, kernel="IBS")
coxKM(Z=Z, U=phenotype2$time, Delta=phenotype2$event, X=X, gamma=Gamma, kernel="linear")

izhbannikov/vartools documentation built on May 17, 2017, 5:29 a.m.