RDkit＆mol2vec ：靶标抑制剂活性二分类模型对比

pip install git+https://github.com/samoturk/mol2vec

#导入依赖库
import numpy as npimport pandas as pdimport seaborn as snsimport matplotlib.pyplot as pltfrom rdkit import Chem from rdkit.Chem import Drawfrom rdkit.Chem import Descriptorsimport warningswarnings.filterwarnings("ignore")%matplotlib inline

#载入数据，并查看data = pd.read_csv('HIV.csv')data.head()

#查看靶标数目分布sns.countplot(data = data, x='HIV_active', orient='v')plt.ylabel('HIM active')plt.xlabel('Count of values')plt.show()

#SMILES 转 MOLdata['mol'] = data['smiles'].apply(lambda x: Chem.MolFromSmiles(x)) 
#计算分子描述符data['tpsa'] = data['mol'].apply(lambda x: Descriptors.TPSA(x))data['mol_w'] = data['mol'].apply(lambda x: Descriptors.ExactMolWt(x))data['num_valence_electrons'] = data['mol'].apply(lambda x: Descriptors.NumValenceElectrons(x))data['num_heteroatoms'] = data['mol'].apply(lambda x: Descriptors.NumHeteroatoms(x))

#预处理数据并划分数据集from sklearn.model_selection import train_test_split
y = data.HIV_active.valuesX = data.drop(columns=['smiles', 'activity','HIV_active', 'mol'])
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.20, random_state=1)

#定义ROC 和 AUC 绘图函数from sklearn.metrics import auc, roc_curvedef evaluation_class(model, X_test, y_test):    prediction = model.predict_proba(X_test)    preds = model.predict_proba(X_test)[:,1]    fpr, tpr, threshold = roc_curve(y_test, preds)    roc_auc = auc(fpr, tpr)
    plt.title('ROC Curve')    plt.plot(fpr, tpr, 'g', label = 'AUC = %0.3f' % roc_auc)    plt.legend(loc = 'lower right')    plt.plot([0, 1], [0, 1],'r--')    plt.xlim([0, 1])    plt.ylim([0, 1])    plt.ylabel('True Positive Rate')    plt.xlabel('False Positive Rate')    plt.show()
    print('ROC AUC score:', round(roc_auc, 4))

#建立逻辑回归预测模型，且结果绘图from sklearn.linear_model import LogisticRegressionfrom sklearn.preprocessing import StandardScaler
X_train = StandardScaler().fit_transform(X_train)X_test = StandardScaler().fit_transform(X_test)
lr = LogisticRegression()lr.fit(X_train, y_train)
evaluation_class(lr, X_test, y_test)

#加载预先经过word2vec训练的模型from gensim.models import word2vecmodel = word2vec.Word2Vec.load('model_300dim.pkl')

#导入mol2vec，计算分子矢量特征，训练模型且结果绘图
from mol2vec.features import mol2alt_sentence, mol2sentence, MolSentence, DfVec, sentences2vecfrom gensim.models import word2vec
#Constructing sentencesdata['sentence'] = data.apply(lambda x: MolSentence(mol2alt_sentence(x['mol'], 1)), axis=1)
#Extracting embeddings to a numpy.array#Note that we always should mark unseen='UNK' in sentence2vec() so that model is taught how to handle unknown substructuresdata['mol2vec'] = [DfVec(x) for x in sentences2vec(data['sentence'], model, unseen='UNK')]X_mol = np.array([x.vec for x in data['mol2vec']])X_mol = pd.DataFrame(X_mol)
#Concatenating matrices of featuresnew_data = pd.concat((X, X_mol), axis=1)
X_train, X_test, y_train, y_test = train_test_split(new_data, y, test_size=.20, random_state=1)
X_train = StandardScaler().fit_transform(X_train)X_test = StandardScaler().fit_transform(X_test)
lr = LogisticRegression()lr.fit(X_train, y_train)

evaluation_class(lr, X_test, y_test)