rm(list = ls())
############################################library packages
#install.packages("keras")
library(deepnet)
library(h2o)
## Warning: package 'h2o' was built under R version 4.2.3
##
## ----------------------------------------------------------------------
##
## Your next step is to start H2O:
## > h2o.init()
##
## For H2O package documentation, ask for help:
## > ??h2o
##
## After starting H2O, you can use the Web UI at http://localhost:54321
## For more information visit https://docs.h2o.ai
##
## ----------------------------------------------------------------------
##
## Attaching package: 'h2o'
## The following objects are masked from 'package:stats':
##
## cor, sd, var
## The following objects are masked from 'package:base':
##
## %*%, %in%, &&, ||, apply, as.factor, as.numeric, colnames,
## colnames<-, ifelse, is.character, is.factor, is.numeric, log,
## log10, log1p, log2, round, signif, trunc
library(neuralnet)
## Warning: package 'neuralnet' was built under R version 4.2.3
library(keras)
## Warning: package 'keras' was built under R version 4.2.3
library(tensorflow)
## Warning: package 'tensorflow' was built under R version 4.2.3
library(tidyverse)
## ── Attaching packages
## ───────────────────────────────────────
## tidyverse 1.3.2 ──
## ✔ ggplot2 3.4.0 ✔ purrr 1.0.1
## ✔ tibble 3.1.8 ✔ dplyr 1.0.10
## ✔ tidyr 1.3.0 ✔ stringr 1.5.0
## ✔ readr 2.1.3 ✔ forcats 0.5.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::compute() masks neuralnet::compute()
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
library(splitstackshape)
library(ROCR)
##
## Attaching package: 'ROCR'
##
## The following object is masked from 'package:neuralnet':
##
## prediction
library(pROC)
## Type 'citation("pROC")' for a citation.
##
## Attaching package: 'pROC'
##
## The following object is masked from 'package:h2o':
##
## var
##
## The following objects are masked from 'package:stats':
##
## cov, smooth, var
library(mltools)
##
## Attaching package: 'mltools'
##
## The following object is masked from 'package:tidyr':
##
## replace_na
##########################################prepare data path
dir_path_ori <- "C:\\Users\\xut2\\Desktop\\keras\\"
fs_dir <- c("auc_feature_selection")
for (ii in 1:length(fs_dir)) {
#ii = 1
dir_path <- paste0(dir_path_ori,fs_dir[ii],"\\selected_data_for_model")
dir_path_target <- list.dirs(dir_path,recursive = F)
dir_path_target
for (nn in 1:length(dir_path_target)) {
tryCatch({
#nn= 1
###creat folders for save results
unlink(paste0(dir_path_target[nn],"\\","deepnet_test"), recursive=TRUE) #delete file and all involved
unlink(paste0(dir_path_target[nn],"\\","h2o_test"), recursive=TRUE) #delete file and all involved
unlink(paste0(dir_path_target[nn],"\\","neuralnet_test"), recursive=TRUE) #delete file and all involved
unlink(paste0(dir_path_target[nn],"\\","keras_test"), recursive=TRUE) #delete file and all involved
dir_path_name <- dir(dir_path_target[nn],pattern = "*.csv") ##input data
dir.create(paste0(dir_path_target[nn],"\\","deepnet_test"))
dir.create(paste0(dir_path_target[nn],"\\","h2o_test"))
dir.create(paste0(dir_path_target[nn],"\\","neuralnet_test"))
dir.create(paste0(dir_path_target[nn],"\\","keras_test"))
##########################################bath input data
#aa=1
object_file_list <- list()
for (aa in 1:length(dir_path_name)) {
dir_path_file <- paste0(dir_path_target[nn],"\\",dir_path_name[aa])
object_file_list[[aa]] <- read.csv(dir_path_file,
check.names = T, header = T,stringsAsFactors = F)
names(object_file_list)[aa] <- dir_path_name[aa]
}
#names(object_file_list)
#############################################################model
for (mm in 1:length(object_file_list)) {
#mm=1
#################################################data for model
data <- object_file_list[[mm]] #check point- 1
#dim(data) #[1] 208 26
#colnames(data)
colnames(data) <- c(colnames(data[1:(ncol(data)-1)]),"num")
data[,-c(1)] <- sapply(data[,-c(1)], as.numeric)###################################################ADD__1
#data <- data.frame(sapply(data, as.numeric))
data$num <- as.factor(data$num)
print(str(data)); print(head(data, 3)); print(dim(data)); print(colnames(data))
###############################################parameter setting
loop <- 2 # Repeat 10 times
n_col_newdata <- ncol(data)-1
n_col_newdata
###deepnet
auc_deepnet <- numeric(loop)
accuracy_deepnet <- numeric(loop)
Balanced_accuracy_deepnet <- numeric(loop)
mcc_result_deepnet <- numeric(loop)
###h2o
auc_h2o <- numeric(loop)
accuracy_h2o <- numeric(loop)
Balanced_accuracy_h2o <- numeric(loop)
mcc_result_h2o <- numeric(loop)
###neuralnet
auc_neuralnet <- numeric(loop)
accuracy_neuralnet <- numeric(loop)
Balanced_accuracy_neuralnet <- numeric(loop)
mcc_result_neuralnet <- numeric(loop)
###keras
auc_keras <- numeric(loop)
accuracy_keras <- numeric(loop)
Balanced_accuracy_keras <- numeric(loop)
mcc_result_keras <- numeric(loop)
###############################################divide data
#loop = 2
for (j in 1:loop) {
tryCatch({
set.seed(j*2023)
#j =1
#####################################################################
#####################################################################
data_test_list <- stratified(data, "num", .3)$Mapping.ID
#data_test_list <- sample(unique(data$Mapping.ID),size = floor(length(unique(data$Mapping.ID))*0.3),replace = F)
# Randomly shuffle the data
#k <- 5 # 5 fold cross validation #check point
testdata <- data[data$Mapping.ID %in% data_test_list,]
traindata <- data[!data$Mapping.ID %in% data_test_list,]
#dim(testdata);dim(traindata) #[1] 3840 36 [1] 8992 36
#unique(testdata$Mapping.ID) %in% unique(traindata$Mapping.ID)
testdata$Mapping.ID <- NULL
traindata$Mapping.ID <- NULL
testdata <- unique(testdata)
traindata <- unique(traindata)
#str(traindata);str(testdata)
#print(head(traindata))
#unique(traindata$num)
#table(traindata$num) #0 1 461 311
#dim(testdata);dim(traindata) [1] 525 34 [1] 793 34
#####################################################################
#############################################deepnet
set.seed(2023)
fit_deepnet <- nn.train(as.matrix(traindata[, -ncol(traindata)]),
as.numeric(as.character(traindata[, ncol(traindata)])),
hidden = c(ceiling(c(ncol(traindata)-1)/2),ceiling(c(ncol(traindata)-1)/3)))
#fit_deepnet[1:10]
############################
# Predict the test set probabilities
#For classification task,return the probability of a class
pre_auc_deepnet <- nn.predict(fit_deepnet, testdata[, -ncol(testdata)])
#### auc_deepnet
modelroc_deepnet <- pROC::roc(testdata[,n_col_newdata], as.numeric(pre_auc_deepnet[,1]))
auc_deepnet[j] <- as.numeric(auc(modelroc_deepnet))
cutoff_deepnet <- coords(modelroc_deepnet, "best", ret = "threshold")$threshold[1]
predict.results_deepnet <- ifelse(pre_auc_deepnet[,1] > as.numeric(cutoff_deepnet[1]),"1","0")
freq_default_deepnet <- table(predict.results_deepnet, testdata[,n_col_newdata])
Sensitivity_deepnet <- freq_default_deepnet[1,1]/sum(freq_default_deepnet[,1])
Specificity_deepnet <- freq_default_deepnet[2,2]/sum(freq_default_deepnet[,2])
accuracy_deepnet[j] <- mean(testdata[,n_col_newdata] == predict.results_deepnet)
Balanced_accuracy_deepnet[j] <- (Sensitivity_deepnet+Specificity_deepnet)/2
###mcc
preds_deepnet <- as.numeric(as.character(predict.results_deepnet))
actuals_deepnet <- as.numeric(as.character(testdata$num))
mcc_result_deepnet[j] <- mcc(actuals_deepnet,preds_deepnet)
mcc_result_deepnet[j] <- abs(mcc_result_deepnet[j])
##
if (auc_deepnet[j] < 0.5) {auc_deepnet[j] = 1 - auc_deepnet[j]
accuracy_deepnet[j] <- 1- accuracy_deepnet[j]
Balanced_accuracy_deepnet[j] <- 1- Balanced_accuracy_deepnet[j]
}
print(data.frame(auc_deepnet[j], accuracy_deepnet[j], Balanced_accuracy_deepnet[j], mcc_result_deepnet[j]))
},error=function(e){cat("ERROR :",conditionMessage(e),"\n")})
############################################## h2o
tryCatch({
h2o.init()
traindata_h2o <- as.h2o(traindata)
testdata_h2o <- as.h2o(testdata)
colnames(traindata_h2o)
#str(traindata_h2o)
################################build model
#?h2o.deeplearning
#?h2o.deeplearning
#set.seed(2023)
h2o_fit <- h2o.deeplearning(
x = 1:c(ncol(testdata_h2o)-1),
y = ncol(testdata_h2o),
training_frame = traindata_h2o,
hidden = c(ceiling(c(ncol(traindata)-1)/2),ceiling(c(ncol(traindata)-1)/3)),
seed = 2023, reproducible = TRUE
)
# Predict the test set probabilities
#For classification task,return the probability of a class
pred_prob_h2o <- h2o.predict(h2o_fit, testdata_h2o[, -ncol(testdata_h2o)], exact_predict = TRUE)
roc_curve_h2o <- roc(as.numeric(as.matrix(testdata_h2o$num)), as.numeric(as.matrix(pred_prob_h2o$p1)),
plot = TRUE, print.auc=TRUE)
roc_curve_h2o
#### auc_h2o
auc_h2o[j] <- as.numeric(auc(roc_curve_h2o))
####Balanced accuracy_h2o_method_1
modelroc_h2o <- coords(roc_curve_h2o, "best", ret = "threshold")$threshold[1]
#class(modelroc_h2o)
predict.results_h2o <- ifelse(pred_prob_h2o[, grep("p1", colnames(pred_prob_h2o))] > as.numeric(modelroc_h2o[1]),"1","0")
freq_default_h2o <- table(as.numeric(as.matrix(predict.results_h2o)),as.numeric(as.matrix(testdata_h2o[,n_col_newdata])))
Sensitivity_h2o <- freq_default_h2o[1,1]/sum(freq_default_h2o[,1])
Specificity_h2o <- freq_default_h2o[2,2]/sum(freq_default_h2o[,2])
accuracy_h2o[j] <- mean(testdata_h2o[,n_col_newdata] == predict.results_h2o)
Balanced_accuracy_h2o[j] <- (Sensitivity_h2o+Specificity_h2o)/2
###mcc
preds_h2o <- as.numeric(as.matrix(predict.results_h2o))
actuals_h2o <- as.numeric(as.matrix(testdata_h2o$num))
mcc_result_h2o[j] <- mcc(actuals_h2o,preds_h2o)
mcc_result_h2o[j] <- abs(mcc_result_h2o[j])
##
if (auc_h2o[j] < 0.5) {auc_h2o[j] = 1 - auc_h2o[j]
accuracy_h2o[j] <- 1- accuracy_h2o[j]
Balanced_accuracy_h2o[j] <- 1- Balanced_accuracy_h2o[j]
}
#h2o.shutdown(prompt = F)
print(data.frame(auc_h2o[j], accuracy_h2o[j],Balanced_accuracy_h2o[j], mcc_result_h2o[j]))
},error=function(e){cat("ERROR :",conditionMessage(e),"\n")})
############################################## neuralnet
tryCatch({
n <- names(traindata)
f <- as.formula(paste("num ~", paste(n[!n %in% "num"], collapse = " + ")))
traindata_nn <- traindata
testdata_nn <- testdata
traindata_nn$num <- as.numeric(as.character(traindata_nn$num))
testdata_nn$num <- as.numeric(as.character(testdata_nn$num))
colnames(traindata_nn)
str(traindata_nn)
set.seed(2023)
#traindata_nn[, -ncol(traindata_nn)] <- sapply(traindata_nn[, -ncol(traindata_nn)], scale)
nn_fit <- neuralnet(f, data = traindata_nn, linear.output = FALSE,
hidden = c(ceiling(c(length(n)-1)/2),ceiling(c(length(n)-1)/3)))
#plot(nn_fit)
#?neuralnet
############################
# Predict the test set probabilities
#For classification task,return the probability of a class
#str(testdata[, -ncol(testdata)])
pre_auc_neuralnet <- predict(nn_fit, testdata_nn[, -ncol(testdata_nn)])
roc_curve_nn <- roc(testdata_nn$num, pre_auc_neuralnet[,1], plot = TRUE, print.auc=TRUE)
roc_curve_nn
#neuralnet::compute(nn_fit, testdata[, -ncol(testdata)])$net.result
#### auc_neuralnet
auc_neuralnet[j] <- as.numeric(auc(roc_curve_nn))
####Balanced accuracy_neuralnet_method_1
modelroc_neuralnet <- roc_curve_nn
cutoff_neuralnet <- coords(modelroc_neuralnet, "best", ret = "threshold")$threshold[1]
predict.results_neuralnet <- ifelse(pre_auc_neuralnet[,1] > as.numeric(cutoff_neuralnet[1]),"1","0")
freq_default_neuralnet <- table(predict.results_neuralnet,testdata[,n_col_newdata])
Sensitivity_neuralnet <- freq_default_neuralnet[1,1]/sum(freq_default_neuralnet[,1])
Specificity_neuralnet <- freq_default_neuralnet[2,2]/sum(freq_default_neuralnet[,2])
accuracy_neuralnet[j] <- mean(testdata[,n_col_newdata] == predict.results_neuralnet)
Balanced_accuracy_neuralnet[j] <- (Sensitivity_neuralnet+Specificity_neuralnet)/2
###mcc
preds_neuralnet <- as.numeric(as.character(predict.results_neuralnet))
actuals_neuralnet <- as.numeric(as.character(testdata$num))
mcc_result_neuralnet[j] <- mcc(actuals_neuralnet,preds_neuralnet)
if (auc_neuralnet[j] < 0.5) {auc_neuralnet[j] = 1 - auc_neuralnet[j]
accuracy_neuralnet[j] <- 1- accuracy_neuralnet[j]
Balanced_accuracy_neuralnet[j] <- 1- Balanced_accuracy_neuralnet[j]
mcc_result_neuralnet[j] <- abs(mcc_result_neuralnet[j])}
print(data.frame(auc_neuralnet[j], accuracy_neuralnet[j],
Balanced_accuracy_neuralnet[j], mcc_result_neuralnet[j]))
},error=function(e){cat("ERROR :",conditionMessage(e),"\n")})
##############################################keras
tryCatch({
#c(ceiling(c(ncol(traindata)-1)/3),ceiling(c(ncol(traindata)-1)/4))
model <- keras_model_sequential()
model %>%
layer_dense(units = ceiling(c(ncol(traindata)-1)/2), activation = 'relu', input_shape = ncol(traindata)-1) %>%
layer_dropout(rate = 0.4) %>%
layer_dense(units = ceiling(c(ncol(traindata)-1)/3), activation = 'relu') %>%
layer_dropout(rate = 0.3) %>%
#layer_dense(units = 64, activation = 'relu') %>%
#layer_dropout(rate = 0.2) %>%
layer_dense(units = 1, activation = 'sigmoid')
################################Before fitting the model, we compile it specifying the type of loss function,
#the optimizer and the metric used for evaluating performance.
model %>% compile(
loss = 'binary_crossentropy',
optimizer = "adam",
metrics = c('AUC'))
summary(model)
################################We use fit to estimate network parameters.
str(traindata)
traindata_keras <- traindata
testdata_keras <- testdata
traindata_keras$num <- as.numeric(as.character(traindata_keras$num))
testdata_keras$num <- as.numeric(as.character(testdata_keras$num))
str(traindata_keras)
history <- model %>% fit(
x = as.matrix(traindata_keras[, -ncol(traindata_keras)]),
y = as.matrix(traindata_keras[, ncol(traindata_keras)]),
epochs = 20,
batch_size = 32,
validation_split = 0.2,
verbose=0)
plot(history)
#############################Evaluate the model’s performance on the test data:
model %>% evaluate(as.matrix(traindata_keras[, -ncol(traindata_keras)]),
as.matrix(traindata_keras[, ncol(traindata_keras)]), batch_size=32)
model %>% evaluate(as.matrix(testdata_keras[, -ncol(testdata_keras)]),
as.matrix(testdata_keras[, ncol(testdata_keras)]), batch_size=32) #0.7960
############################
# Predict the test set probabilities
#print(auc_keras)
#### auc_keras
pre_auc_keras <- predict(model, as.matrix(testdata_keras[, -ncol(testdata_keras)]), type = "prob")
pred_auc_keras <- prediction(pre_auc_keras[,1],testdata_keras[,n_col_newdata])
auc_keras[j] <- performance(pred_auc_keras,"auc")@y.values[[1]]
####Balanced accuracy_keras_method_1
modelroc_keras <- pROC::roc(testdata_keras[,n_col_newdata],as.numeric(pre_auc_keras[,1]))
cutoff_keras <- coords(modelroc_keras, "best", ret = "threshold")$threshold[1]
predict.results_keras <- ifelse(pre_auc_keras[,1] > as.numeric(cutoff_keras[1]),"1","0")
freq_default_keras <- table(predict.results_keras,testdata[,n_col_newdata])
Sensitivity_keras <- freq_default_keras[1,1]/sum(freq_default_keras[,1])
Specificity_keras <- freq_default_keras[2,2]/sum(freq_default_keras[,2])
accuracy_keras[j] <- mean(testdata[,n_col_newdata] == predict.results_keras)
Balanced_accuracy_keras[j] <- (Sensitivity_keras+Specificity_keras)/2
###mcc
preds_keras <- as.numeric(as.character(predict.results_keras))
actuals_keras <- as.numeric(as.character(testdata$num))
mcc_result_keras[j] <- mcc(actuals_keras,preds_keras)
##
if (auc_keras[j] < 0.5) {auc_keras[j] = 1 - auc_keras[j]
accuracy_keras[j] <- 1- accuracy_keras[j]
Balanced_accuracy_keras[j] <- 1- Balanced_accuracy_keras[j]
mcc_result_keras[j] <- abs(mcc_result_keras[j])}
print(data.frame(auc_keras[j], accuracy_keras[j],
Balanced_accuracy_keras[j], mcc_result_keras[j]))
},error=function(e){cat("ERROR :",conditionMessage(e),"\n")})
}
####################################output data
#####deepnet
output_deepnet <- data.frame(mean(auc_deepnet[auc_deepnet != 0]),sd(auc_deepnet[auc_deepnet != 0]),
mean(accuracy_deepnet[accuracy_deepnet != 0]),sd(accuracy_deepnet[accuracy_deepnet != 0]),
mean(Balanced_accuracy_deepnet[Balanced_accuracy_deepnet != 0]),sd(Balanced_accuracy_deepnet[Balanced_accuracy_deepnet != 0]),
mean(mcc_result_deepnet[mcc_result_deepnet != 0]),sd(mcc_result_deepnet[mcc_result_deepnet != 0]))
file_name_data_deepnet <- paste0(dir_path_target[nn],"\\","deepnet_test","\\","deepnet_test_",names(object_file_list)[[mm]])
write.csv(output_deepnet,file_name_data_deepnet)
#####h2o
output_h2o <- data.frame(mean(auc_h2o[auc_h2o != 0]),sd(auc_h2o[auc_h2o != 0]),
mean(accuracy_h2o[accuracy_h2o != 0]),sd(accuracy_h2o[accuracy_h2o != 0]),
mean(Balanced_accuracy_h2o[Balanced_accuracy_h2o != 0]),sd(Balanced_accuracy_h2o[Balanced_accuracy_h2o != 0]),
mean(mcc_result_h2o[mcc_result_h2o != 0]),sd(mcc_result_h2o[mcc_result_h2o != 0]))
file_name_data_h2o <- paste0(dir_path_target[nn],"\\","h2o_test","\\","h2o_test_",names(object_file_list)[[mm]])
write.csv(output_h2o,file_name_data_h2o)
#####neuralnet
output_neuralnet <- data.frame(mean(auc_neuralnet[auc_neuralnet != 0]),sd(auc_neuralnet[auc_neuralnet != 0]),
mean(accuracy_neuralnet[accuracy_neuralnet != 0]),sd(accuracy_neuralnet[accuracy_neuralnet != 0]),
mean(Balanced_accuracy_neuralnet[Balanced_accuracy_neuralnet != 0]),sd(Balanced_accuracy_neuralnet[Balanced_accuracy_neuralnet != 0]),
mean(mcc_result_neuralnet[mcc_result_neuralnet != 0]),sd(mcc_result_neuralnet[mcc_result_neuralnet != 0]))
file_name_data_neuralnet <- paste0(dir_path_target[nn],"\\","neuralnet_test","\\","neuralnet_test_",names(object_file_list)[[mm]])
write.csv(output_neuralnet,file_name_data_neuralnet)
#####keras
output_keras <- data.frame(mean(auc_keras[auc_keras != 0]),sd(auc_keras[auc_keras != 0]),
mean(accuracy_keras[accuracy_keras != 0]),sd(accuracy_keras[accuracy_keras != 0]),
mean(Balanced_accuracy_keras[Balanced_accuracy_keras != 0]),sd(Balanced_accuracy_keras[Balanced_accuracy_keras != 0]),
mean(mcc_result_keras[mcc_result_keras != 0]),sd(mcc_result_keras[mcc_result_keras != 0]))
file_name_data_keras <- paste0(dir_path_target[nn],"\\","keras_test","\\","keras_test_",names(object_file_list)[[mm]])
write.csv(output_keras,file_name_data_keras)
}
},error=function(e){cat("ERROR :",conditionMessage(e),"\n")})
}
#sink()
}
## 'data.frame': 929 obs. of 22 variables:
## $ Mapping.ID : chr "AAKJLRGGTJKAMG" "AAOVKJBEBIDNHE" "ABJKWBDEJIDSJZ" "ACGDKVXYNVEAGU" ...
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 0 1 0 0 0 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 1 1 0 0 0 1 1 1 1 1 ...
## $ tox21.erb.bla.p1_ch2 : num 0 1 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 0 0 1 1 1 1 1 0 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 1 1 0 1 0 1 1 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 1 1 0 0 0 1 1 0 0 1 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 0 1 1 0 0 0 1 1 0 1 ...
## $ tox21.ahr.p1_ratio : num 1 0 0 0 1 0 0 0 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 1 1 0 0 1 1 0 0 1 1 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 1 1 0 0 0 0 0 1 1 0 ...
## $ tox21.are.bla.p1_ch2 : num 1 0 0 0 1 1 0 1 0 1 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 1 0 0 0 0 1 0 0 1 0 ...
## $ tox21.mitotox.p1_rhodamine : num 1 0 0 1 0 0 0 0 0 0 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 1 0 1 1 0 0 0 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 0 0 0 0 0 1 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 0 1 1 0 1 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 1 1 0 0 0 1 0 0 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 1 0 1 1 0 0 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 0 1 1 1 1 1 1 0 1 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 1 0 0 0 0 0 1 1 ...
## $ num : Factor w/ 2 levels "0","1": 2 1 2 1 2 2 2 2 2 2 ...
## NULL
## Mapping.ID tox21.mitotox.p1_ratio tox21.adrb2.agonist.p1_ratio.curve.rank
## 1 AAKJLRGGTJKAMG 1 1
## 2 AAOVKJBEBIDNHE 0 1
## 3 ABJKWBDEJIDSJZ 1 0
## tox21.erb.bla.p1_ch2 tox21.gr.hela.bla.antagonist.p1_ch2
## 1 0 0
## 2 1 1
## 3 0 0
## tox21.p450.2c9.p1_ratio.curve.rank tox21.chrm1.p1_antagonist.curve.rank
## 1 1 1
## 2 1 1
## 3 1 0
## tox21.p450.1a2.p1_ratio.curve.rank tox21.ahr.p1_ratio
## 1 0 1
## 2 1 0
## 3 1 0
## tox21.adrb2.agonist.p1_ch1.curve.rank tox21.ppard.bla.antagonist.p1_ch2
## 1 1 1
## 2 1 1
## 3 0 0
## tox21.are.bla.p1_ch2 tox21.vdr.bla.antagonist.p1_ch2
## 1 1 1
## 2 0 0
## 3 0 0
## tox21.mitotox.p1_rhodamine tox21.shh.3t3.gli3.antagonist.p1_ratio
## 1 1 1
## 2 0 1
## 3 0 1
## tox21.kiss1r.hek293.p2_antagonist.curve.rank
## 1 0
## 2 0
## 3 0
## tox21.er.luc.bg1.4e2.agonist.p2_ratio tox21.p450.3a4.p1_ratio.curve.rank
## 1 0 1
## 2 1 1
## 3 0 0
## tox21.gnrhr.hek293.p2_antagonist.curve.rank tox21.pparg.bla.antagonist.p1_ch2
## 1 1 0
## 2 1 1
## 3 1 1
## tox21.htr2a.p1_antagonist.curve.rank num
## 1 1 1
## 2 0 0
## 3 1 1
## [1] 929 22
## [1] "Mapping.ID"
## [2] "tox21.mitotox.p1_ratio"
## [3] "tox21.adrb2.agonist.p1_ratio.curve.rank"
## [4] "tox21.erb.bla.p1_ch2"
## [5] "tox21.gr.hela.bla.antagonist.p1_ch2"
## [6] "tox21.p450.2c9.p1_ratio.curve.rank"
## [7] "tox21.chrm1.p1_antagonist.curve.rank"
## [8] "tox21.p450.1a2.p1_ratio.curve.rank"
## [9] "tox21.ahr.p1_ratio"
## [10] "tox21.adrb2.agonist.p1_ch1.curve.rank"
## [11] "tox21.ppard.bla.antagonist.p1_ch2"
## [12] "tox21.are.bla.p1_ch2"
## [13] "tox21.vdr.bla.antagonist.p1_ch2"
## [14] "tox21.mitotox.p1_rhodamine"
## [15] "tox21.shh.3t3.gli3.antagonist.p1_ratio"
## [16] "tox21.kiss1r.hek293.p2_antagonist.curve.rank"
## [17] "tox21.er.luc.bg1.4e2.agonist.p2_ratio"
## [18] "tox21.p450.3a4.p1_ratio.curve.rank"
## [19] "tox21.gnrhr.hek293.p2_antagonist.curve.rank"
## [20] "tox21.pparg.bla.antagonist.p1_ch2"
## [21] "tox21.htr2a.p1_antagonist.curve.rank"
## [22] "num"
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
## auc_deepnet.j. accuracy_deepnet.j. Balanced_accuracy_deepnet.j.
## 1 0.567569 0.5683453 0.5683453
## mcc_result_deepnet.j.
## 1 0.1387749
## Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 55 minutes 49 seconds
## H2O cluster timezone: America/New_York
## H2O data parsing timezone: UTC
## H2O cluster version: 3.40.0.1
## H2O cluster version age: 2 months and 17 days
## H2O cluster name: H2O_started_from_R_xut2_nzr889
## H2O cluster total nodes: 1
## H2O cluster total memory: 0.94 GB
## H2O cluster total cores: 20
## H2O cluster allowed cores: 20
## H2O cluster healthy: TRUE
## H2O Connection ip: localhost
## H2O Connection port: 54321
## H2O Connection proxy: NA
## H2O Internal Security: FALSE
## R Version: R version 4.2.2 (2022-10-31 ucrt)
##
##
|
| | 0%
|
|======================================================================| 100%
##
|
| | 0%
|
|======================================================================| 100%
##
|
| | 0%
|
|======================================================================| 100%
##
|
| | 0%
|
|======================================================================| 100%
## Setting levels: control = 0, case = 1
## Setting direction: controls > cases
## auc_h2o.j. accuracy_h2o.j. Balanced_accuracy_h2o.j. mcc_result_h2o.j.
## 1 0.5190725 0.4496403 0.4496403 0.1499923
## 'data.frame': 642 obs. of 21 variables:
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 1 0 0 0 1 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 1 1 0 0 1 1 1 1 1 0 ...
## $ tox21.erb.bla.p1_ch2 : num 0 1 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 0 1 1 1 1 1 0 1 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 1 1 1 0 1 1 0 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 1 1 0 0 1 1 0 0 1 0 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 0 1 1 0 0 1 1 0 1 0 ...
## $ tox21.ahr.p1_ratio : num 1 0 0 1 0 0 0 0 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 1 1 0 1 1 0 0 1 1 1 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 1 1 0 0 0 0 1 1 0 1 ...
## $ tox21.are.bla.p1_ch2 : num 1 0 0 1 1 0 1 0 1 0 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 1 0 0 0 1 0 0 1 0 0 ...
## $ tox21.mitotox.p1_rhodamine : num 1 0 0 0 0 0 0 0 0 1 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 0 1 1 0 0 0 1 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 0 0 0 0 1 1 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 1 1 0 1 0 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 1 1 0 0 1 0 0 0 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 0 1 1 0 0 1 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 0 1 1 1 1 1 0 1 0 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 1 0 0 0 0 1 1 0 ...
## $ num : num 1 0 1 1 1 1 1 1 1 0 ...
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
## auc_neuralnet.j. accuracy_neuralnet.j. Balanced_accuracy_neuralnet.j.
## 1 0.5293981 0.5539568 0.5539568
## mcc_result_neuralnet.j.
## 1 0.1205724
## Model: "sequential"
## ________________________________________________________________________________
## Layer (type) Output Shape Param #
## ================================================================================
## dense_2 (Dense) (None, 10) 210
## dropout_1 (Dropout) (None, 10) 0
## dense_1 (Dense) (None, 7) 77
## dropout (Dropout) (None, 7) 0
## dense (Dense) (None, 1) 8
## ================================================================================
## Total params: 295
## Trainable params: 295
## Non-trainable params: 0
## ________________________________________________________________________________
## 'data.frame': 642 obs. of 21 variables:
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 1 0 0 0 1 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 1 1 0 0 1 1 1 1 1 0 ...
## $ tox21.erb.bla.p1_ch2 : num 0 1 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 0 1 1 1 1 1 0 1 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 1 1 1 0 1 1 0 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 1 1 0 0 1 1 0 0 1 0 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 0 1 1 0 0 1 1 0 1 0 ...
## $ tox21.ahr.p1_ratio : num 1 0 0 1 0 0 0 0 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 1 1 0 1 1 0 0 1 1 1 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 1 1 0 0 0 0 1 1 0 1 ...
## $ tox21.are.bla.p1_ch2 : num 1 0 0 1 1 0 1 0 1 0 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 1 0 0 0 1 0 0 1 0 0 ...
## $ tox21.mitotox.p1_rhodamine : num 1 0 0 0 0 0 0 0 0 1 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 0 1 1 0 0 0 1 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 0 0 0 0 1 1 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 1 1 0 1 0 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 1 1 0 0 1 0 0 0 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 0 1 1 0 0 1 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 0 1 1 1 1 1 0 1 0 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 1 0 0 0 0 1 1 0 ...
## $ num : Factor w/ 2 levels "0","1": 2 1 2 2 2 2 2 2 2 1 ...
## 'data.frame': 642 obs. of 21 variables:
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 1 0 0 0 1 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 1 1 0 0 1 1 1 1 1 0 ...
## $ tox21.erb.bla.p1_ch2 : num 0 1 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 0 1 1 1 1 1 0 1 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 1 1 1 0 1 1 0 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 1 1 0 0 1 1 0 0 1 0 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 0 1 1 0 0 1 1 0 1 0 ...
## $ tox21.ahr.p1_ratio : num 1 0 0 1 0 0 0 0 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 1 1 0 1 1 0 0 1 1 1 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 1 1 0 0 0 0 1 1 0 1 ...
## $ tox21.are.bla.p1_ch2 : num 1 0 0 1 1 0 1 0 1 0 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 1 0 0 0 1 0 0 1 0 0 ...
## $ tox21.mitotox.p1_rhodamine : num 1 0 0 0 0 0 0 0 0 1 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 0 1 1 0 0 0 1 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 0 0 0 0 1 1 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 1 1 0 1 0 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 1 1 0 0 1 0 0 0 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 0 1 1 0 0 1 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 0 1 1 1 1 1 0 1 0 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 1 0 0 0 0 1 1 0 ...
## $ num : num 1 0 1 1 1 1 1 1 1 0 ...
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
## auc_keras.j. accuracy_keras.j. Balanced_accuracy_keras.j. mcc_result_keras.j.
## 1 0.5583044 0.5611511 0.5611511 0.1223528
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
## auc_deepnet.j. accuracy_deepnet.j. Balanced_accuracy_deepnet.j.
## 1 0.5177277 0.5395683 0.5415631
## mcc_result_deepnet.j.
## 1 0.09984338
## Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 56 minutes 51 seconds
## H2O cluster timezone: America/New_York
## H2O data parsing timezone: UTC
## H2O cluster version: 3.40.0.1
## H2O cluster version age: 2 months and 17 days
## H2O cluster name: H2O_started_from_R_xut2_nzr889
## H2O cluster total nodes: 1
## H2O cluster total memory: 0.20 GB
## H2O cluster total cores: 20
## H2O cluster allowed cores: 20
## H2O cluster healthy: TRUE
## H2O Connection ip: localhost
## H2O Connection port: 54321
## H2O Connection proxy: NA
## H2O Internal Security: FALSE
## R Version: R version 4.2.2 (2022-10-31 ucrt)
##
##
|
| | 0%
|
|======================================================================| 100%
##
|
| | 0%
|
|======================================================================| 100%
##
|
| | 0%
|
|======================================================================| 100%
##
|
| | 0%
|
|======================================================================| 100%
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
## auc_h2o.j. accuracy_h2o.j. Balanced_accuracy_h2o.j. mcc_result_h2o.j.
## 1 0.5107919 0.5359712 0.5348344 0.07344363
## 'data.frame': 643 obs. of 21 variables:
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 1 1 1 0 0 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 0 1 1 1 1 0 1 1 1 0 ...
## $ tox21.erb.bla.p1_ch2 : num 0 0 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 1 1 0 0 1 1 1 1 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 0 1 1 0 0 1 1 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 0 1 1 0 1 0 0 1 0 0 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 1 0 1 1 1 0 1 1 0 1 ...
## $ tox21.ahr.p1_ratio : num 0 0 0 0 0 0 0 1 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 0 1 0 0 1 0 1 1 1 0 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 0 0 0 1 0 0 1 1 1 0 ...
## $ tox21.are.bla.p1_ch2 : num 0 1 0 1 1 0 1 1 0 1 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 0 1 0 0 0 0 0 1 0 0 ...
## $ tox21.mitotox.p1_rhodamine : num 0 0 0 0 0 1 0 1 0 0 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 0 0 1 1 1 1 0 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 1 0 1 0 0 0 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 1 0 0 0 1 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 0 1 0 0 0 1 1 1 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 0 1 1 1 1 0 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 1 1 1 0 0 0 1 0 1 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 0 0 1 0 0 1 0 0 ...
## $ num : num 1 1 1 1 1 1 1 1 0 0 ...
## Warning: Algorithm did not converge in 1 of 1 repetition(s) within the stepmax.
## ERROR : requires numeric/complex matrix/vector arguments
## Model: "sequential_1"
## ________________________________________________________________________________
## Layer (type) Output Shape Param #
## ================================================================================
## dense_5 (Dense) (None, 10) 210
## dropout_3 (Dropout) (None, 10) 0
## dense_4 (Dense) (None, 7) 77
## dropout_2 (Dropout) (None, 7) 0
## dense_3 (Dense) (None, 1) 8
## ================================================================================
## Total params: 295
## Trainable params: 295
## Non-trainable params: 0
## ________________________________________________________________________________
## 'data.frame': 643 obs. of 21 variables:
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 1 1 1 0 0 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 0 1 1 1 1 0 1 1 1 0 ...
## $ tox21.erb.bla.p1_ch2 : num 0 0 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 1 1 0 0 1 1 1 1 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 0 1 1 0 0 1 1 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 0 1 1 0 1 0 0 1 0 0 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 1 0 1 1 1 0 1 1 0 1 ...
## $ tox21.ahr.p1_ratio : num 0 0 0 0 0 0 0 1 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 0 1 0 0 1 0 1 1 1 0 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 0 0 0 1 0 0 1 1 1 0 ...
## $ tox21.are.bla.p1_ch2 : num 0 1 0 1 1 0 1 1 0 1 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 0 1 0 0 0 0 0 1 0 0 ...
## $ tox21.mitotox.p1_rhodamine : num 0 0 0 0 0 1 0 1 0 0 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 0 0 1 1 1 1 0 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 1 0 1 0 0 0 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 1 0 0 0 1 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 0 1 0 0 0 1 1 1 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 0 1 1 1 1 0 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 1 1 1 0 0 0 1 0 1 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 0 0 1 0 0 1 0 0 ...
## $ num : Factor w/ 2 levels "0","1": 2 2 2 2 2 2 2 2 1 1 ...
## 'data.frame': 643 obs. of 21 variables:
## $ tox21.mitotox.p1_ratio : num 1 0 1 0 0 1 1 1 0 0 ...
## $ tox21.adrb2.agonist.p1_ratio.curve.rank : num 0 1 1 1 1 0 1 1 1 0 ...
## $ tox21.erb.bla.p1_ch2 : num 0 0 0 0 0 0 0 0 0 0 ...
## $ tox21.gr.hela.bla.antagonist.p1_ch2 : num 0 1 1 1 0 0 1 1 1 1 ...
## $ tox21.p450.2c9.p1_ratio.curve.rank : num 1 0 1 1 0 0 1 1 0 0 ...
## $ tox21.chrm1.p1_antagonist.curve.rank : num 0 1 1 0 1 0 0 1 0 0 ...
## $ tox21.p450.1a2.p1_ratio.curve.rank : num 1 0 1 1 1 0 1 1 0 1 ...
## $ tox21.ahr.p1_ratio : num 0 0 0 0 0 0 0 1 0 0 ...
## $ tox21.adrb2.agonist.p1_ch1.curve.rank : num 0 1 0 0 1 0 1 1 1 0 ...
## $ tox21.ppard.bla.antagonist.p1_ch2 : num 0 0 0 1 0 0 1 1 1 0 ...
## $ tox21.are.bla.p1_ch2 : num 0 1 0 1 1 0 1 1 0 1 ...
## $ tox21.vdr.bla.antagonist.p1_ch2 : num 0 1 0 0 0 0 0 1 0 0 ...
## $ tox21.mitotox.p1_rhodamine : num 0 0 0 0 0 1 0 1 0 0 ...
## $ tox21.shh.3t3.gli3.antagonist.p1_ratio : num 1 1 1 0 0 1 1 1 1 0 ...
## $ tox21.kiss1r.hek293.p2_antagonist.curve.rank: num 0 0 0 0 1 0 1 0 0 0 ...
## $ tox21.er.luc.bg1.4e2.agonist.p2_ratio : num 0 1 0 1 0 0 0 1 0 0 ...
## $ tox21.p450.3a4.p1_ratio.curve.rank : num 0 1 0 0 0 1 1 1 0 0 ...
## $ tox21.gnrhr.hek293.p2_antagonist.curve.rank : num 1 1 1 0 1 1 1 1 0 1 ...
## $ tox21.pparg.bla.antagonist.p1_ch2 : num 1 1 1 0 0 0 1 0 1 0 ...
## $ tox21.htr2a.p1_antagonist.curve.rank : num 1 0 0 0 1 0 0 1 0 0 ...
## $ num : num 1 1 1 1 1 1 1 1 0 0 ...
## Setting levels: control = 0, case = 1
## Setting direction: controls > cases
## auc_keras.j. accuracy_keras.j. Balanced_accuracy_keras.j. mcc_result_keras.j.
## 1 0.5114648 0.5395683 0.5375776 0.09026935