MSENet_Beta: Estimate beta_hat using MSE loss function along with Net...

In xueweic/APGD: TGPred: Efficient methods for predicting target genes of a transcription factor by integrating statistics, machine learning, and optimization.

Estimate beta_hat using MSE loss function along with Net penalty function

Usage

MSENet_Beta(X, y, Adj, lambda0, alpha0, method = "APGD", gamma = 1000, niter = 2000, crit_beta = 1e-04, crit_obj = 1e-08, quiet = FALSE, if.scale=FALSE)

Arguments

X	expressional levels of n_genes target genes (TGs)
y	expressional levels of a transcription factor (TF)
Adj	the adjacency matrix of network structure.
lambda0	one of parameters in Net regression, which controls the number of nonzero coefficients.
alpha0	one of parameters in Net regression, which controls the numerical values of nonzero coefficients.
method	The current methods must be 'APGD' or 'CVX'
gamma	initial value of gamma in APGD. default: 1000
niter	the maximum number of APGD to solve HuberNet regression. default: 2000
crit_beta	converge criterion of change of beta. default: 1e-4
crit_obj	converge criterion of change of objective function. default: 1e-8
quiet	decide if exist the output report. default: FALSE
if.scale	if.scale decide if scale the expression levels. default: FALSE

Value

beta	n_genes length vector of estimated regulated effect sizes, where beta_j != 0 indicates j th gene is not selected in Net regression.

Examples

lambda0 = 200
  
alpha0 = 0.5
  
beta_hat_APGD <- MSENet_Beta(X, y, Adj, lambda0, alpha0,method="APGD", if.scale=TRUE)
  
library("CVXR")
  
beta_hat_CVX <- MSENet_Beta(X, y, Adj,lambda0, alpha0,method="CVX", if.scale=TRUE)

xueweic/APGD documentation built on Sept. 4, 2023, 2:18 a.m.