Regularized Network-Based Variable Selection
Network-based regularization has achieved success in variable selection for high-dimensional biological data due to its ability to incorporate correlations among genomic features. This package provides procedures of network-based variable selection for generalized linear models (Ren et al.(2017) and Ren et al.(2019)). Continuous, binary, and survival response are supported. Robust network-based methods are available for continuous and survival responses.
install.packages("pak")
pak::pak("jrenstat/regnet")
install.packages("regnet")
The cross-validation examples below use an explicit foldid vector. Set
a seed before running the examples if you want the elastic-net
initialization to be exactly reproducible.
library(regnet)
data(SurvExample)
X = rgn.surv$X
Y = rgn.surv$Y
clv = c(1:5) # variable 1 to 5 are clinical variables, we choose not to penalize them here.
foldid = rep(1:5, length.out=nrow(X))
out = cv.regnet(X, Y, response="survival", penalty="network",
clv=clv, robust=TRUE, foldid=foldid)
out$lambda
fit = regnet(out, out$lambda[1,1], out$lambda[1,2])
index = which(rgn.surv$beta[-(1:6)] != 0) # [-(1:6)] removes the intercept and clinical variables that are not subject to selection.
pos = which(fit$coeff[-(1:6)] != 0)
tp = length(intersect(index, pos))
fp = length(pos) - tp
list(tp=tp, fp=fp)
library(regnet)
data(LogisticExample)
X = rgn.logi$X
Y = rgn.logi$Y
foldid = rep(1:5, length.out=nrow(X))
out = cv.regnet(X, Y, response="binary", penalty="network",
foldid=foldid, r = 4.5, robust=FALSE)
out$lambda
fit = regnet(X, Y, "binary", "network", out$lambda[1,1], out$lambda[1,2], r = 4.5)
index = which(rgn.logi$beta[-1] != 0) # [-1] removes the intercept
pos = which(fit$coeff[-1] != 0)
tp = length(intersect(index, pos))
fp = length(pos) - tp
list(tp=tp, fp=fp)
# Fitting from a cv.regnet object
fit = regnet(out, out$lambda[1,1], out$lambda[1,2])
list(response=fit$para$response,
penalty=fit$para$penalty,
selected=sum(fit$coeff[-1] != 0))
library(regnet)
data(ContExample)
X = rgn.tcga$X
Y = rgn.tcga$Y
clv = (1:2)
fit = regnet(X, Y, "continuous", "network", rgn.tcga$lamb1, rgn.tcga$lamb2, clv =clv, alpha.i=0.5, robust=FALSE)
net = plot(fit)
subs = plot(fit, subnetworks = TRUE, vsize=20, labelDist = 3, theta = 5)
foldid to cv.regnet() for user-controlled cross-validation
folds.regnet() from a cv.regnet object
and explicitly supplied lambda values.QRWMR.Based on users’ feedback, we have
This package provides implementation for methods proposed in
Ren, J., He, T., Li, Y., Liu, S., Du, Y., Jiang, Y., Wu, C. (2017). Network-based regularization for high dimensional SNP data in the case-control study of Type 2 diabetes. BMC Genetics, 18(1):44
Ren, J., Du, Y., Li, S., Ma, S., Jiang,Y. and Wu, C. (2019). Robust network-based regularization and variable selection for high dimensional genomics data in cancer prognosis. Genet. Epidemiol. 43:276-291
Wu, C., and Ma, S. (2015). A selective review of robust variable selection with applications in bioinformatics. Briefings in Bioinformatics, 16(5), 873–883
Wu, C., Shi, X., Cui, Y. and Ma, S. (2015). A penalized robust semiparametric approach for gene-environment interactions. Statistics in Medicine, 34 (30): 4016–4030
Wu, C, Jiang, Y, Ren, J, Cui, Y, Ma, S. (2018). Dissecting gene-environment interactions: A penalized robust approach accounting for hierarchical structures.Statistics in Medicine, 37:437–456