-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathautomate.py
173 lines (146 loc) · 6.43 KB
/
automate.py
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
# Import all the necessary modules
import seaborn as sns
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import tensorflow as tf
import pickle
from sonnia.processing import Processing
from tqdm.notebook import tqdm
from sklearn.preprocessing import LabelEncoder
from sklearn.preprocessing import OneHotEncoder
from sklearn.cluster import KMeans
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Input
from tensorflow.keras.layers import Dense
from tensorflow.keras.layers import ReLU
from tensorflow.keras.layers import BatchNormalization
from tensorflow.keras.utils import plot_model
from tensorflow.keras.layers import Concatenate
from sklearn.decomposition import PCA
def train_model_basic(df, X_train, params={}):
if isinstance(X_train, list):
X_train = X_train[0]
# define encoder
visible = Input(shape=(params["n_inputs"],))
e = Dense(params["n_inputs"]*2)(visible)
e = BatchNormalization()(e)
e = ReLU()(e)
# define bottleneck
n_bottleneck = 2
bottleneck = Dense(n_bottleneck)(e)
# define decoder
d = Dense(params["n_inputs"]*2)(bottleneck)
d = BatchNormalization()(d)
d = ReLU()(d)
# output layer
output = Dense(params["n_inputs"], activation='linear')(d)
# define autoencoder model
model = Model(inputs=visible, outputs=output)
# compile autoencoder model
model.compile(optimizer='adam', loss='mse')
# compile autoencoder model
model.compile(optimizer='adam', loss='mse')
# fit the autoencoder model to reconstruct input
history = model.fit(X_train, X_train, epochs=params["epochs"], batch_size=params["batch_size"], verbose=1)
# Return encoder part of the model
encoder = Model(inputs=visible, outputs=bottleneck)
#Plot the history loss of the model
plt.plot(history.history['loss'], label='train')
plt.title('Model loss')
plt.legend()
plt.show()
return model, encoder
def train_model_complex(df, X_train, params={}):
# define encoder
cdr3_input = Input(shape=(params["n_inputs"],), name='cdr3_input')
v_gene_input = Input(shape=(params["v_inputs"],), name='v_gene_input')
j_gene_input = Input(shape=(params["j_inputs"],), name='j_gene_input')
cdr3_embedding = Dense(params["n_inputs"]*2)(cdr3_input)
cdr3_embedding = BatchNormalization()(cdr3_embedding)
cdr3_embedding = ReLU()(cdr3_embedding)
v_gene_embedding = Dense(params['v_gene_embedding_dim'], name='v_gene_embedding')(v_gene_input)
j_gene_embedding = Dense(params['j_gene_embedding_dim'], name='j_gene_embedding')(j_gene_input)
merged_embedding = Concatenate(axis=1,name='merged_embedding')([cdr3_embedding, v_gene_embedding, j_gene_embedding])
encoder_dense_1 = Dense(params['dense_nodes'], activation='elu', name='encoder_dense_1')(merged_embedding)
encoder_dense_2 = Dense(params['dense_nodes'], activation='elu', name='encoder_dense_2')(encoder_dense_1)
# Latent layers:
bottleneck = Dense(params['latent_dim'], name='bottleneck')(encoder_dense_2)
# define decoder
d = Dense(params["n_inputs"]*2)(bottleneck)
d = BatchNormalization()(d)
d = ReLU()(d)
# output layer
output = Dense(params["n_inputs"], activation='linear')(d)
# define autoencoder model
model = Model([cdr3_input, v_gene_input, j_gene_input], [output])
# compile autoencoder model
model.compile(optimizer='adam', loss='mse')
# compile autoencoder model
model.compile(optimizer='adam', loss='mse')
print(model.summary())
# fit the autoencoder model to reconstruct input
history = model.fit(X_train, X_train, epochs=params["epochs"], batch_size=64, verbose=1)
# Return encoder part of the model
encoder = Model(inputs=[cdr3_input, v_gene_input, j_gene_input], outputs=bottleneck)
decoder = Model(inputs=bottleneck, outputs=output)
#Plot the history loss of the model
plt.plot(history.history['loss'][10:], label='complex model', color='blue')
plt.title('Model loss')
plt.legend()
plt.show()
return model, encoder, decoder
def model_results(df, X_test, encoder, label="label", mode_complex=True):
if label == "end_seq_label":
df[label] = df["CDR3_al"].apply(lambda x: x[15:])
elif label == "begin_seq_label":
df[label] = df["CDR3_al"].apply(lambda x: x[:5])
elif label == "j_gene":
df[label] = df["j_gene"].apply(lambda x: x.split("-")[0])
labels = []
labels_encoder = LabelEncoder()
labels_encoder = labels_encoder.fit(df[label].unique())
for k in tqdm(df.index):
labels.append(labels_encoder.transform([df.loc[k,label]]))
labels = [int(y) for y in labels]
df[label] = labels
rgb_values = sns.color_palette("Spectral", df[label].nunique())
df[str(label+"_color")] = df[label].apply(lambda x: rgb_values[x])
N = 300
if mode_complex:
X_test = X_test.copy()
for i in range(len(X_test)):
X_test[i] = X_test[i][:N]
X_test_encode = encoder.predict(X_test[:N])
pca = PCA(n_components=2)
principalComponents = pca.fit_transform(X_test_encode)
plot_X_test = principalComponents
else:
if isinstance(X_test, list):
X_test = X_test[0]
X_test_encode = encoder.predict(X_test[:N])
plot_X_test = X_test_encode.copy()
print(len(X_test_encode))
print(X_test_encode.shape)
plt.scatter(plot_X_test[:,0], plot_X_test[:,1], color=df[str(label+"_color")][:N], alpha=0.5)
plt.title(label)
plt.show()
k = 8
kmeans = KMeans(n_clusters=k, random_state=0).fit(X_test_encode[:N])
labels = pd.Series(kmeans.labels_)
rgb_values = sns.color_palette("Spectral", k)
col_kmeans = labels.apply(lambda x: rgb_values[x])
sample = df[:N]
sample[str(label+"_kmeans_label")] = kmeans.labels_
label_dict = {}
for cluster in range(k):
label_ind = sample[sample[str(label+"_kmeans_label")] == cluster][label].value_counts().index[0]
label_dict[cluster] = label_ind
print("Accuracy: ", sum([label_dict[x] == y for x,y in zip(sample[str(label+"_kmeans_label")], sample[label])])/len(sample))
plt.scatter(plot_X_test[:,0], plot_X_test[:,1], color=col_kmeans, alpha=0.5)
centroids = kmeans.cluster_centers_
if not mode_complex:
for cluster in range(k):
plt.text(centroids[cluster,0], centroids[cluster,1], labels_encoder.inverse_transform([label_dict[cluster]]), fontsize=10)
plt.title("Kmeans clustering for "+label)
plt.show()