python sklearn svm svr 多输出_RDKit | 基于支持向量回归(SVR)预测logP

RDKit一个用于化学信息学的python库。使用支持向量回归(SVR)来预测logP。分子的输入结构特征是摩根指纹，输出是logP。

代码示例：

1. 导入依赖库

2. import numpy as npfrom rdkit import Chemfrom rdkit.Chem.Crippen import MolLogPfrom rdkit import Chem, DataStructsfrom rdkit.Chem import AllChemfrom sklearn.svm import SVRfrom sklearn.metrics import mean_squared_error, r2_scorefrom scipy import statsimport matplotlib.pyplot as plt
              

载入smile分子库，计算morgan指纹和logP

num_mols = 5000f = open('smiles.txt', 'r')contents = f.readlines()fps_total = []logP_total = []for i in range(num_mols):smi = contents[i].split()[0]m = Chem.MolFromSmiles(smi)fp = AllChem.GetMorganFingerprintAsBitVect(m,2)arr = np.zeros((1,))DataStructs.ConvertToNumpyArray(fp,arr)fps_total.append(arr)logP_total.append(MolLogP(m))fps_total = np.asarray(fps_total)logP_total = np.asarray(logP_total)
           

划分训练集和测试集

num_total = fps_total.shape[0]num_train = int(num_total*0.8)num_total, num_train, (num_total-num_train)fps_train = fps_total[0:num_train]logP_train = logP_total[0:num_train]fps_test = fps_total[num_train:]logP_test = logP_total[num_train:]
           

将SVR模型用于回归模型

https://scikit-learn.org/stable/modules/generated/sklearn.svm.SVR.html

1. _gamma = 5.0clf = SVR(kernel='poly', gamma=_gamma)clf.fit(fps_train, logP_train)
              

完成训练后，应该检查预测的准确性。对于评估，将使用r2和指标的均方误差。

1. logP_pred = clf.predict(fps_test)r2 = r2_score(logP_test, logP_pred)mse = mean_squared_error(logP_test, logP_pred)r2, mse
              

模型结果可视化

1. slope, intercept, r_value, p_value, std_error = stats.linregress(logP_test, logP_pred)yy = slope*logP_test+interceptplt.scatter(logP_test, logP_pred, color='black', s=1)plt.plot(logP_test, yy, label='Predicted logP = '+str(round(slope,2))+'*True logP + '+str(round(intercept,2)))plt.xlabel('True logP')plt.ylabel('Predicted logP')plt.legend()plt.show()
              

参考

https://github.com/SeongokRyu/CH485---Artificial-Intelligence-and-Chemistry

https://blog.csdn.net/zb123455445/article/details/78354489