-
Notifications
You must be signed in to change notification settings - Fork 0
/
ML_algorithms.R
191 lines (165 loc) · 7.56 KB
/
ML_algorithms.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
#REQUIREMENTS
library("igraph")
library("dplyr")
library("ggplot2")
library("tidyr")
library("enrichR")
library("plyr")
#set pathway
setwd("~/R-4.2.1/predict_complexes_ML")
options(scipen=999)
#import_files
stats_pdbs_net<-read.csv("pdbs_net.csv", fileEncoding = "UTF-8-BOM", header=TRUE)
positive_cases<-read.csv("final_positive_cases.csv", fileEncoding = "UTF-8-BOM", header=TRUE)
#all_possible_triangles<-readRDS("onen_triangles_net_new_ids.RData")
edges_net_hippie<-read.csv("edges_new_ids.csv", fileEncoding = "UTF-8-BOM", header=TRUE)
nodes_net_hippie<-read.csv("features_nodes_new_ids.csv", fileEncoding = "UTF-8-BOM", header=TRUE)
ML_data<-read.csv("randomized_v1v2.csv", fileEncoding = "UTF-8-BOM", header=TRUE)
#statistics/plots
hist_pdb_num_prot_net<-ggplot(stats_pdbs_net, aes(x = number_proteins)) +
geom_histogram(binwidth = 1, fill = "#F8766D", color = "black", alpha = 1) +
labs(x = "Number of proteins/PDB",
y = "Frequency")+theme_bw()+
scale_x_continuous(breaks = seq(3,52, by = 2))
hist_pdb_num_prot_net
hist_pdb_num_ot_pdb<-ggplot(stats_pdbs_net, aes(x = number_ot)) +
geom_histogram(binwidth = 14, fill = "#F8766D", color = "black", alpha = 7) +
labs(x = "Number of open triangles",
y = "Frequency")+theme_bw()
hist_pdb_num_ot_pdb
pdb_stats<-ggarrange(hist_pdb_num_prot_net,hist_pdb_num_ot_pdb,
ncol = 2,labels = c("A", "B"), nrow = 1)
pdb_stats
#ML analysis
library("randomForest")
require(caTools)
library('caret')
library("ROSE")
library("pROC")
df_ml<-ML_data%>%
dplyr::select(1,5:46) %>%
tidyr::drop_na()
summary(df_ml)
names(df_ml) <- make.names(names(df_ml))
df_ml$positive<-as.factor(df_ml$positive)
table(df_ml$positive)
#split in train and test set
set.seed(452016)
sample_df_ml = sample.split(df_ml$positive, SplitRatio = .70)
train_df_ml = subset(df_ml, sample_df_ml == TRUE)
test_df_ml = subset(df_ml, sample_df_ml == FALSE)
dim(train_df_ml)
dim(test_df_ml)
table(train_df_ml$positive)
#undersampling
data_balanced_under <- ovun.sample(positive~.,data = train_df_ml,method = "under",N=810)$data
data_balanced_under_replace <- data_balanced_under %>%
dplyr::mutate(positive = ifelse(positive == 0,"No","Yes"))
test_df_ml_replace<-test_df_ml %>%
dplyr::mutate(positive = ifelse(positive == 0,"No","Yes"))
#random forest
rf_under<- train(positive ~ ., data = data_balanced_under,
method = 'rf',metric="Accuracy",
trControl=trainControl(method="repeatedcv",
repeats = 10,
number=5,
returnResamp='final',savePredictions = TRUE))
rf_under
pred_class_rf= predict(rf_under,newdata=test_df_ml[-1],type='raw')
#pred_class_rf_prob= predict(rf_under,newdata=test_df_ml[-1],type='prob')
cm_rf = as.data.frame(table(test_df_ml[,1], as.numeric(pred_class_rf)))
cm_rf<-caret::confusionMatrix(pred_class_rf,as.factor(test_df_ml$positive),
mode="everything")
cm_rf
#SVM
svm_under<- train(positive ~ ., data = data_balanced_under_replace,
method = "svmLinear",metric="Accuracy",
trControl=trainControl(method="repeatedcv",
repeats = 10,
number=5,returnResamp='final',
savePredictions = TRUE,classProbs = TRUE))
svm_under
pred_test_svm_class= predict(svm_under,newdata=test_df_ml_replace[-1],type='raw')
#pred_test_svm_prob= predict(svm_under,newdata=test_df_ml_replace[-1],type='prob')
cm_svm = as.data.frame(table(test_df_ml_replace[,1], as.numeric(pred_test_svm_class)))
cm_svm<-caret::confusionMatrix(pred_test_svm_class,as.factor(test_df_ml_replace$positive),
mode="everything")
cm_svm
#logistic regression
lg_under<- train(positive ~ ., data = data_balanced_under,
method = 'glmnet',metric="Accuracy",
trControl=trainControl(method="repeatedcv",
repeats = 10,
number=5,
returnResamp='final',savePredictions = TRUE))
lg_under
pred_class_lg= predict(lg_under,newdata=test_df_ml[-1],type='raw')
#pred_prob_lg= predict(lg_under,newdata=test_df_ml[-1],type='prob')
cm_lg = as.data.frame(table(test_df_ml[,1], as.numeric(pred_class_lg)))
cm_lg<-caret::confusionMatrix(pred_class_lg,as.factor(test_df_ml$positive),
mode="everything")
cm_lg
#decission_tree
dt_under<- train(positive ~ ., data = data_balanced_under,
method = 'rpart',metric="Accuracy",
trControl=trainControl(method="repeatedcv",
repeats = 10,
number=5,
returnResamp='final',savePredictions = TRUE))
dt_under
pred_class_dt= predict(dt_under,newdata=test_df_ml[-1],type='raw')
#pred_prob_dt= predict(dt_under,newdata=test_df_ml[-1],type='prob')
cm_dt = as.data.frame(table(test_df_ml[,1], as.numeric(pred_class_dt)))
cm_dt<-caret::confusionMatrix(pred_class_dt,as.factor(test_df_ml$positive),
mode="everything")
cm_dt
#knn
knn_under<- train(positive ~ ., data = data_balanced_under,
method = 'knn',metric="Accuracy",
trControl=trainControl(method="repeatedcv",
repeats = 10,
number=5,
returnResamp='final',savePredictions = TRUE))
knn_under
pred_class_knn= predict(knn_under,newdata=test_df_ml[-1],type='raw')
#pred_prob_knn= predict(knn_under,newdata=test_df_ml[-1],type='prob')
cm_knn = as.data.frame(table(test_df_ml[,1], as.numeric(pred_class_knn)))
cm_knn<-caret::confusionMatrix(pred_class_knn,as.factor(test_df_ml$positive),
mode="everything")
cm_knn
#apply the model to the dataset
predictions_triplets= predict(rf_under,newdata=df_ml,type='prob')
final_predictions_triplets<-cbind(ML_data,predictions_triplets)
final_predictions<-final_predictions_triplets%>%
dplyr::select(1:4,48)%>%
dplyr::rename("score"=5)
cooperative<-final_predictions[final_predictions$score >= 0.9,]
competitive<-final_predictions[final_predictions$score <= 0.1,]
hist(final_predictions$score)
final_predictions_top_score<-final_predictions[final_predictions$score >= 0.5,]
#feature importance
i_scores_caret <- varImp(rf_under,scale = FALSE)
print(i_scores_caret)
feat_imp<-plot(i_scores_caret)
feat_imp
importance_df<-as.data.frame(i_scores_caret$importance)
importance_df$feature <- rownames(importance_df)
rownames(importance_df) <- c()
importance_df_or<-importance_df[order(-importance_df$Overall),]
library("forcats")
feat_impo<-ggplot(importance_df_or, aes(x=fct_reorder(feature, Overall),y=Overall)) +
geom_bar(stat="identity",color="black",fill="#00BFC4",alpha=0.7,width = 0.6)+
theme_bw()+coord_flip()+xlab("")+
ylab("Feature importance")
feat_impo
#roc curve
library("pROC")
roc_score=roc(test_df_ml[,5], as.numeric(pred_down_inside[[2]]))
auc <- roc_score$auc
roc_plot<-ggroc(roc_score, colour = '#00BFC4', size = 0.8,legacy.axes = T) +
theme_bw()+
geom_segment(aes(x = 0, xend = 1, y = 0, yend = 1),color="darkgrey")
roc_plot
plot_model<-ggarrange(accu_mod,roc_plot,
ncol = 2,labels = c("A", "B"), nrow = 1)
plot_model