-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathHistone_code.r
192 lines (177 loc) · 5.31 KB
/
Histone_code.r
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
library(glmnet)
library(Matrix)
library(foreach)
library(pROC)
library(caret)
library(lattice)
# Input data
data <- data <- read.csv("./data/example_histone.csv",header=T)
histone_name <- colnames(data)
# Generate the nonlinear feature library
record_2 <- c()
x <- 1:5
for (i in 1:5) {
for (j in i:5) {
a <- c(x[i],x[j])
record_2 <-rbind(record_2,a)
}
}
record_3 <- c()
x <- 1:5
for (i in 1:5) {
for (j in i:5) {
for (k in j:5) {
a <- c(x[i],x[j],x[k])
record_3 <-rbind(record_3,a)}
}
}
record_4 <- c()
x<-1:5
for (i in 1:5) {
for (j in i:5) {
for (k in j:5) {
for (m in k:5) {
a <- c(x[i],x[j],x[k],x[m])
record_4 <- rbind(record_4,a)
}
}
}
}
record_5 <- c()
x<-1:5
for (i in 1:5) {
for (j in i:5) {
for (k in j:5) {
for (m in k:5) {
for (n in m:5) {
a <- c(x[i],x[j],x[k],x[m],x[n])
record_5 <- rbind(record_5,a)
}
}
}
}
}
# Combination of two histone features
combin_matrix <- t(record_2)
record <- matrix(NA,nrow(data),ncol(combin_matrix))
record_name <- c()
for(i in 1:ncol(combin_matrix)){
feature_vec <- combin_matrix[,i]
data_sub = data[,feature_vec[1]]*data[,feature_vec[2]]
sub_name = histone_name[feature_vec]
merge_name<- paste(sub_name[1],"*",sub_name[2],sep = "")
record[,i] <- data_sub
record_name[i] <- merge_name
}
colnames(record)<-record_name
histone_2 <- record
# Combination of three histone features
combin_matrix <- t(record_3)
record <- matrix(NA,nrow(data),ncol(combin_matrix))
record_name <- c()
for(i in 1:ncol(combin_matrix)){
feature_vec <- combin_matrix[,i]
data_sub = data[,feature_vec[1]]*data[,feature_vec[2]]*data[,feature_vec[3]]
sub_name = histone_name[feature_vec]
merge_name<- paste(sub_name[1],"*",sub_name[2],"*",sub_name[3],sep = "")
record[,i] <- data_sub
record_name[i] <- merge_name
}
colnames(record)<-record_name
histone_3 <- record
# Combination of four histone features
combin_matrix <- t(record_4)
record <- matrix(NA,nrow(data),ncol(combin_matrix))
record_name <- c()
for(i in 1:ncol(combin_matrix)){
feature_vec <- combin_matrix[,i]
data_sub = data[,feature_vec[1]]*data[,feature_vec[2]]*data[,feature_vec[3]]*data[,feature_vec[4]]
sub_name = histone_name[feature_vec]
merge_name<- paste(sub_name[1],"*",sub_name[2],"*",sub_name[3],"*",sub_name[4],sep = "")
record[,i] <- data_sub
record_name[i] <- merge_name
}
colnames(record)<-record_name
histone_4 <- record
# Combination of five histone features
combin_matrix <- t(record_5)
record <- matrix(NA,nrow(data),ncol(combin_matrix))
record_name <- c()
for(i in 1:ncol(combin_matrix)){
feature_vec <- combin_matrix[,i]
data_sub = data[,feature_vec[1]]*data[,feature_vec[2]]*data[,feature_vec[3]]*data[,feature_vec[4]]*data[,feature_vec[5]]
sub_name = histone_name[feature_vec]
merge_name<- paste(sub_name[1],"*",sub_name[2],"*",sub_name[3],"*",sub_name[4],sub_name[5],sep = "")
record[,i] <- data_sub
record_name[i] <- merge_name
}
colnames(record)<-record_name
histone_5 <- record
data <- cbind(histone_2,histone_3,histone_4,histone_5,data)
# Ordinary logistic regression
OLR <- function(data, folds = 5) {
set.seed(2)
require(caret)
all_fold_predict <- c()
all_true_value <- c()
for (i in 1:folds) {
fold_test <- data[createFolds(y = data$Label, k = folds)[[i]],]
fold_train <- data[-createFolds(y = data$Label, k = folds)[[i]],]
fold_pre <- glm(Label ~ ., family = binomial(link = 'logit'), data = fold_train)
fold_predict <- predict(fold_pre, type = 'response', newdata = fold_test)
true_value <- fold_test$Label
all_fold_predict <- c(all_fold_predict, fold_predict)
all_true_value <- c(all_true_value, true_value)
}
modelroc <- roc(all_true_value, all_fold_predict)
auc_value <- auc(modelroc)
auc_value<-round(auc_value,4)
return(auc_value)
}
# Evaluate individual histone prediction performance
feature_record <- c()
auc_record<-c()
feature <- data[,-ncol(data)]
feature_name <- colnames(feature)
feature_length <- length(feature_name)
record <- matrix(NA,feature_length,2)
for (i in 1:feature_length) {
print(i)
data_sub <- data.frame(feature[,i],data$Label)
colnames(data_sub) <- c(feature_name[i],"Label")
auc_value <- OLR(data_sub)
record[i,1] <- feature_name[i]
record[i,2] <- auc_value
}
value <- as.numeric(record[,2])
index <- which(value==max(value))
feature_base <- feature[,index]
feature_all <- feature[,-(index)]
feature_record[1]<- feature_name[index]
auc_record[1] <- max(value)
# Forward feature search
steps = 30
for (k in 2:steps) {
feature <- feature_all
feature_name <- colnames(feature)
feature_length <- length(feature_name)
record <- matrix(NA,feature_length,2)
for (i in 1:feature_length) {
data_sub <- data.frame(feature_base,feature[,i],data$Label)
colnames(data_sub) <- c(feature_record,feature_name[i],"Label")
auc_value <- OLR(data_sub)
record[i,1] <- feature_name[i]
record[i,2] <- auc_value
}
value <- as.numeric(record[,2])
index <- which(value==max(value))
feature_base <- cbind(feature_base,feature[,index])
feature_all <- feature[,-(index)]
feature_record[k]<- feature_name[index]
auc_record[k] <- max(value)
if(auc_record[k]<auc_record[k-1]){
break
}
}
result <- cbind(feature_record,auc_record)
write.csv(feature_record,"output_path/result.csv")