LUCIDus (version 2.1.0)

tune.lucid: Grid search for tuning parameters to fit the LUCID model

Description

Grid search for tuning parameters to fit the LUCID model

Usage

tune.lucid(
  G,
  Z,
  Y,
  CoG = NULL,
  CoY = NULL,
  family = "normal",
  useY = TRUE,
  K = 2:6,
  Rho_G = NULL,
  Rho_Z_InvCov = NULL,
  Rho_Z_CovMu = NULL
)

Arguments

G

Genetic features/environmental exposures, a matrix.

Z

Biomarkers/other omics data, a matrix.

Y

Disease outcome, it is suggested to transform it into a n by 1 matrix.

CoG

Optional, matrix. Covariates to be adjusted for estimating the latent cluster.

CoY

Optional, matrix. Covariates to be adjusted for estimating the outcome.

family

Type of outcome Y. It should be choose from "normal", "binary".

useY

Whether or not to include the information of Y to estimate the latent clusters. Default is TRUE.

K

Numeric sequence. Number of latent clusters.

Rho_G

Numeric sequence, Lasso type penalty for selection of G.

Rho_Z_InvCov

Numeric sequence, Lasso type penalty for the inverse covariance structure of Z.

Rho_Z_CovMu

Numeric sequence, Lasso type penalty for the product of covariance matrix and mean of Z

Value

A list. Containing model BICs of different combination of tuning parameters.

Examples

# NOT RUN {
tuenpar <- tune.lucid(G = G1, Z = Z1, Y = Y1, family = "binary",
Rho_G = seq(0.01, 0.02, by = 0.005),
Rho_Z_InvCov = seq(0.1, 0.3, by = 0.1),
Rho_Z_CovMu = seq(80, 100, by = 10))
# }