KCCA.test: KCCA (Kernel Canonical-Correlation Analysis) based GGI...

Description Usage Arguments Details Value References See Also Examples

View source: R/KCCA.R

Description

KCCA.test performs a Gene-Gene Interaction (GGI) analysis based on the difference of canonical correlations between cases and controls. The "kernel trick" is applied to the canonical correlation to allow the detection of non-linear co-association.

Usage

1
2
3
KCCA.test(Y, G1, G2,kernel=c("rbfdot","polydot","tanhdot","vanilladot","laplacedot",
"besseldot","anovadot","splinedot"),n.boot = 500,sigma=0.05,degree=1,scale=1,offset=1,
order=1)

Arguments

Y

numeric or factor vector with exactly two different values. Y is the response variable and should be of length equal to the number of rows of G1 and G2 arguments (number of individuals).

G1

SnpMatrix object. Must have a number of rows equal to the length of Y.

G2

SnpMatrix object. Must have a number of rows equal to the length of Y.

kernel

A character string matching one of the kernel name in : "rbfdot","polydot","tanhdot","vanilladot","laplacedot","besseldot","anovadot","splinedot". For more details regarding kernel function see dots object. Default is a Gaussian RBF kernel

n.boot

positive integer. n.boot is the number of bootstrap replicates for estimating variances. By default, this is fixed to 500.

sigma

The inverse kernel width used by the Gaussian the Laplacian (kernel="rbfdot"), the Bessel (kernel="besseldot") and the ANOVA (kernel="anovadot") kernel. Default is sigma=0.05.

degree

The degree of the polynomial (kernel="polydot"), bessel (kernel="besseldot") or ANOVA (kernel="anovadot") kernel function. This has to be an positive integer. Default is degree=1.

scale

The scaling parameter of the polynomial (kernel="polydot") and tangent (kernel="tanhdot") kernel is a convenient way of normalizing patterns without the need to modify the data itself. Default is scale=1.

offset

The offset used in a polynomial (kernel="polydot") or hyperbolic tangent (kernel="tanhdot") kernel. Default is offset=1.

order

The order of the Bessel function to be used as a kernel (kernel="besseldot"). Default is order=1.

Details

The test statistic is based on the difference between a Fisher's transformation of the maximum of the kernelized canonical correlations in cases and controls. To calculate the test statistic for the interaction pvalue, KCCA.test estimates the variance of the Fisher's transformation of the maximum of the kernelized canonical correlations in cases and controls using a bootstrap method. The computation of kcca. can be very long.

Value

A list with class "htest" containing the following components:

statistic

The value of the statistic KCCU.

p.value

The p-value for the test.

estimate

A vector of the Fisher's transformed maximum kernel canonical correlation coefficient in cases and controls.

parameter

The number of boostrap samples used to estimate the p-value.

null.value

The value of KCCU under the null hypothesis.

alternative

a character string describing the alternative.

method

a character string indicating the method used.

data.name

a character string giving the names of the data.

References

Yuan, Z. et al. (2012): Detection for gene-gene co-association via kernel canonical correlation analysis, BMC Genetics, 13, 83.
Larson, N. B. et al. (2013): A kernel regression approach to gene-gene interaction detection for case-control studies, Genetic Epidemiology, 37, 695-703.

See Also

GGI, CCA.test

Examples

1
2
3
4
5
data(gene.pair)
## Not run: 
KCCA.test(Y=gene.pair$Y, G1=gene.pair$G1,G2=gene.pair$G2)

## End(Not run)

MathieuEmily/GeneGeneInteR documentation built on Jan. 13, 2018, 6:55 a.m.