Package 'nonnet'

Title: Generate and Analyze Nonlinear Networks
Description: Creates and detects nonlinear relations using the methods described in Slipetz, Qiu, Sun, and Henry (2026) <doi:10.48550/arXiv.2411.02763>. Use the netgen() function to generate a nonlinear network and the dcor_res() function for a residualization procedure for detecting nonlinear relations.
Authors: Lindley Slipetz [aut, cre], Teague Henry [ctb]
Maintainer: Lindley Slipetz <[email protected]>
License: MIT + file LICENSE
Version: 1.0.0
Built: 2026-09-16 12:29:05 UTC
Source: https://github.com/sliplr19/nonnet

Help Index

Distance Correlation Test After Residualization

Description

Residualizes two random variables with respect to a set of covariates using generalized additive models, then tests the association between the residualized variables using a permutation-based distance correlation test.

Usage

dcor_res(dat, A, C, B_variables, nperm)

Arguments

dat

A data frame
A

Character string specifying the name of the first variable to be used in the distance correlation.
C

Character string specifying the name of the second variable to be used in the distance correlation.
B_variables

Character vector containing the names of the covariates used as predictors when residualizing A and C.
nperm

Integer. Number of permutations used in the distance correlation test.

Value

A numeric value giving the permutation p-value from energy::dcor.test().

Examples

dat <- data.frame("A" = c(2,3,4), "C" = c(4,9,16), "B" = c(1,1,1), "D" = c(1,1,1))
res <- dcor_res(dat, "A", "C", c("B", "D"), 1000)
res

Generate Simulated Nonlinear Network Data

Description

Simulates a dataset containing a central variable, peripheral variables, and a target variable under different functional relationships.

Usage

netgen(
  n,
  mean_cent,
  sd_cent,
  mean_peri_non,
  mean_peri_lin,
  sd_peri_non,
  sd_peri_lin,
  mean_target,
  sd_target,
  beta_peri_lin,
  beta_cent_lin,
  beta_lin,
  beta_non,
  beta_cent_non,
  beta_peri_non,
  inter_cent_lin,
  beta_con,
  func,
  con_func
)

Arguments

n

Integer. Number of observations to generate.
mean_cent

Mean of the central variable A.
sd_cent

Standard deviation of the central variable A.
mean_peri_non

Numeric vector. Means of the peripheral nonlinear variables.
mean_peri_lin

Numeric vector. Means of the peripheral linear variables.
sd_peri_non

Numeric vector. Standard deviations of the peripheral nonlinear variables.
sd_peri_lin

Numeric vector. Standard deviations of the peripheral linear variables.
mean_target

Mean of the random error term for the target variable C.
sd_target

Standard deviation of the random error term for the target variable C.
beta_peri_lin

Numeric vector. Regression coefficients for effects of the peripheral linear variables on C.
beta_cent_lin

Numeric vector. Regression coefficients for effects of A on the peripheral linear variables.
beta_lin

Coefficient for the linear effect of A on C.
beta_non

Coefficient for the nonlinear effect of A on C.
beta_cent_non

Numeric vector. Regression coefficients for effects of A on the peripheral nonlinear variables.
beta_peri_non

Numeric vector. Regression coefficients for effects of the peripheral nonlinear variables on C.
inter_cent_lin

Integer vector. Indices identifying peripheral variables involved in interactions with A.
beta_con

Regression coefficient for the confounding effect.
func

Character. Functional form relating A to C.
con_func

Character. Functional form relating A to the nonlinear peripheral variables.

Value

A data frame containing A, C, linear peripheral variables B1, B2, ..., and nonlinear peripheral variables D1, D2, ....

Examples

set.seed(123)

dat <- netgen(
  n = 100,
  mean_cent = 0,
  sd_cent = 1,
  mean_peri_non = rep(0, 4),
  mean_peri_lin = rep(0, 3),
  sd_peri_non = rep(1, 4),
  sd_peri_lin = rep(1, 3),
  mean_target = 0,
  sd_target = 1,
  beta_peri_lin = c(0.3, 0.4, 0.5),
  beta_cent_lin = rep(0.5, 3),
  beta_lin = 0.5,
  beta_non = 0.75,
  beta_cent_non = c(0.5, 0.6, 0.7, 0.8),
  beta_peri_non = c(0.3, 0.4, 0.5, 0.6),
  inter_cent_lin = c(1, 2, 3),
  beta_con = 0,
  func = "quad",
  con_func = "quad"
)

head(dat)