| Type: | Package |
| Title: | New Kernel-Based Test for Differential Association Analysis |
| Version: | 0.1.1 |
| Author: | Hoseung Song [aut, cre], Michael C. Wu [aut] |
| Imports: | mvtnorm |
| Maintainer: | Hoseung Song <hsong3@fredhutch.org> |
| Description: | A new practical method to evaluate whether relationships between two sets of high-dimensional variables are different or not across two conditions. Song, H. and Wu, M.C. (2023) <doi:10.48550/arXiv.2307.15268>. |
| License: | GPL-2 | GPL-3 [expanded from: GPL (≥ 2)] |
| Encoding: | UTF-8 |
| NeedsCompilation: | no |
| Packaged: | 2023-08-22 20:40:23 UTC; hsong3 |
| Repository: | CRAN |
| Date/Publication: | 2023-08-22 21:00:06 UTC |
New kernel-based test for differential association analysis
Description
This package can be used to determine whether two high-dimensional samples have similar dependence relationships across two conditions.
Author(s)
Hoseung Song and Michael C. Wu
Maintainer: Hoseung Song (hsong3@fredhutch.org)
References
Song, H. and Wu, M.C. (2023). Multivariate differential association analysis. arXiv:2307.15268
See Also
Examples
# Dimension of variables.
d = 100
# The first covariance matrix
SIG = matrix(0, d, d)
for (i in 1:d) {
for (j in 1:d) {
SIG[i,j] = 0.4^(abs(i-j))
}
}
# The second covariance matrix
SIG1 = matrix(0, d, d)
for (i in 1:d) {
for (j in 1:d) {
SIG1[i,j] = (0.4+0.5)^(abs(i-j))
}
}
set.seed(500)
# We use 'rmvnorm' in 'mvtnorm' package to generate multivariate normally distributed samples
require(mvtnorm)
Z = rmvnorm(100, mean = rep(0,100), sigma = SIG)
X1 = Z[,1:50]
Y1 = Z[,51:100]
Z = rmvnorm(100, mean = rep(0,100), sigma = SIG1)
X2 = Z[,1:50]
Y2 = Z[,51:100]
a = kerdaa(X1, Y1, X2, Y2, perm=1000)
# output results based on the permutation and the asymptotic results
# the test statistic values can be found in a$stat_g and a$stat_l
# p-values can be found in a$pval and a$pval_perm
New kernel-based test for differential association analysis
Description
This function provides the kernel-based differential association test.
Usage
kerdaa(X1, Y1, X2, Y2, perm=0)
Arguments
X1 |
The first multivariate data in the first condition. |
Y1 |
The second multivariate data in the first condition. |
X2 |
The first multivariate data in the second condition. |
Y2 |
The second multivariate data in the second condition. |
perm |
The number of permutations performed to calculate the p-value of the test. The default value is 0, which means the permutation is not performed and only approximated p-value based on the asymptotic theory is provided. Doing permutation could be time consuming, so be cautious if you want to set this value to be larger than 10,000. |
Value
Returns a list with test statistic values and p-values of the test. See below for more details.
stat_g |
The value of the test statistic using the Gaussian kernel. |
stat_l |
The value of the test statistic using the linear kernel. |
pval |
The omnibus p-value using the approximated p-values of the test statistic based on asymptotic theory. |
pval_perm |
The omnibus p-value using the permutation p-values of the test statistic when argument ‘perm’ is positive. |
See Also
Examples
# Dimension of variables.
d = 100
# The first covariance matrix
SIG = matrix(0, d, d)
for (i in 1:d) {
for (j in 1:d) {
SIG[i,j] = 0.4^(abs(i-j))
}
}
# The second covariance matrix
SIG1 = matrix(0, d, d)
for (i in 1:d) {
for (j in 1:d) {
SIG1[i,j] = (0.4+0.5)^(abs(i-j))
}
}
set.seed(500)
# We use 'rmvnorm' in 'mvtnorm' package to generate multivariate normally distributed samples
require(mvtnorm)
Z = rmvnorm(100, mean = rep(0,100), sigma = SIG)
X1 = Z[,1:50]
Y1 = Z[,51:100]
Z = rmvnorm(100, mean = rep(0,100), sigma = SIG1)
X2 = Z[,1:50]
Y2 = Z[,51:100]
a = kerdaa(X1, Y1, X2, Y2, perm=1000)
# output results based on the permutation and the asymptotic results
# the test statistic values can be found in a$stat_g and a$stat_l
# p-values can be found in a$pval and a$pval_perm