ADCNet/inference.py at main · idrugLab/ADCNet · GitHub

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
import  tensorflow as tf
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from dataset import Graph_Classification_Dataset,Inference_Dataset
from sklearn.metrics import roc_auc_score,confusion_matrix,precision_recall_curve,auc
from rdkit.Chem import Draw
import os
import tensorflow.keras as keras
from model import  PredictModel
import torch
import pickle
from sklearn.preprocessing import StandardScaler
import math

def cover_dict(path):
    file_path = path
    with open(file_path, 'rb') as file:
        data = pickle.load(file)
    tensor_dict = {key: tf.constant(value) for key, value in data.items()}
    new_data = {i: value for i, (key, value) in enumerate(tensor_dict.items())}
    return new_data

def score(y_test, y_pred):
    auc_roc_score = roc_auc_score(y_test, y_pred)
    prec, recall, _ = precision_recall_curve(y_test, y_pred)
    prauc = auc(recall, prec)
    y_pred_print = [round(y, 0) for y in y_pred]
    tn, fp, fn, tp = confusion_matrix(y_test, y_pred_print).ravel()
    se = tp / (tp + fn)
    sp = tn / (tn + fp)  # 也是R
    acc = (tp + tn) / (tp + fn + tn + fp)
    mcc = (tp * tn - fn * fp) / math.sqrt((tp + fn) * (tp + fp) * (tn + fn) * (tn + fp))
    P = tp / (tp + fp)
    F1 = (P * se * 2) / (P + se)
    BA = (se + sp) / 2
    PPV = tp / (tp + fp)
    NPV = tn / (fn + tn)
    return tp, tn, fn, fp, se, sp, mcc, acc, auc_roc_score, F1, BA, prauc, PPV, NPV

def DAR_feature(file_path, column_name):
    df = pd.read_excel(file_path)
    column_data = df[column_name].values.reshape(-1, 1)
    mean_value = 3.86845977
    variance_value = 1.569108443
    std_deviation = variance_value**0.5
    column_data_standardized = (column_data - mean_value) / std_deviation
    normalized_data = (column_data_standardized - 0.8) / (12 - 0.8)
    data_dict = {index: tf.constant(value, dtype=tf.float32) for index, value in zip(df.index, normalized_data.flatten())}
    return data_dict

Heavy_dict = cover_dict('Heavy.pkl')
Light_dict = cover_dict('Light.pkl')
Antigen_dict = cover_dict('Antigen.pkl')
DAR_dict = DAR_feature('data.xlsx', 'DAR')

medium = {'name':'Medium','num_layers': 6, 'num_heads': 8, 'd_model': 256,'path':'medium_weights','addH':True}
arch = medium
trained_epoch = 20
num_layers = arch['num_layers']
num_heads = arch['num_heads']
d_model = arch['d_model']
addH = arch['addH']
dff = d_model * 2
vocab_size = 18
dense_dropout = 0.1
seed = 1
df = pd.read_excel('data.xlsx')
np.random.seed(seed=seed)
tf.random.set_seed(seed=seed)
sml_list1 = df['Payload Isosmiles'].tolist()
sml_list2 = df['Linker Isosmiles'].tolist()

ans = []
y_preds = []
res = []
n = len(sml_list1)
for i in range(n):
    x1 = [sml_list1[i]]
    x2 = [sml_list2[i]]
    t1 = Heavy_dict[i]
    t2 = Light_dict[i]
    t3 = Antigen_dict[i]
    t4 = DAR_dict[i].numpy()
    t1 = tf.expand_dims(t1, axis=0)
    t2 = tf.expand_dims(t2, axis=0)
    t3 = tf.expand_dims(t3, axis=0)
    t4 = tf.constant(t4, shape=(1, 1))

    inference_dataset1 = Inference_Dataset(x1,addH=addH).get_data()
    inference_dataset2 = Inference_Dataset(x2,addH=addH).get_data()

    x1, adjoin_matrix1, smiles1 ,atom_list1 = next(iter(inference_dataset1.take(1)))
    x2, adjoin_matrix2, smiles2 ,atom_list2 = next(iter(inference_dataset2.take(1)))

    seq1 = tf.cast(tf.math.equal(x1, 0), tf.float32)
    seq2 = tf.cast(tf.math.equal(x2, 0), tf.float32)

    mask1 = seq1[:, tf.newaxis, tf.newaxis, :]
    mask2 = seq2[:, tf.newaxis, tf.newaxis, :]

    model = PredictModel(num_layers=num_layers,
                         d_model=d_model,
                         dff=dff,
                         num_heads=num_heads,
                         vocab_size=vocab_size,
                         a=1,
                         dense_dropout = dense_dropout)

    pred = model(x1=x1, mask1=mask1, training=False, adjoin_matrix1=adjoin_matrix1, x2=x2,mask2=mask2, adjoin_matrix2=adjoin_matrix2, t1=t1,t2=t2,t3=t3,t4=t4)
    model.load_weights('classification_weights/ADC_9.h5')

    x = model(x1=x1, mask1=mask1,training=False,adjoin_matrix1=adjoin_matrix1, x2=x2, mask2=mask2, adjoin_matrix2=adjoin_matrix2,t1=t1,t2=t2,t3=t3,t4=t4)
    y_preds.append(x)


y_preds = tf.sigmoid(y_preds)
y_preds = tf.reshape(y_preds,(-1,))
y_hat = tf.where(y_preds < 0.5, 0, 1)
for i in y_preds.numpy():
    ans.append(i)
for i in y_hat.numpy():
    res.append(i)
print(ans)
print(res)