CYPi-DNNpredictor: A Web-server for Prediction of Cytochrome P450 Inhibition Using Deep Neural Network Model


Cytochrome 450 in the human body mediates most drug metabolism, it is very important to screen for adverse drugs in advance and avoid adverse drug interactions during use. CYP 1A2, CYP 2C9, CYP 2C19, CYP 2D6 and CYP 3A4 participates in the metabolism of most drugs in the body. Herein, a python-based deep learning model using the Keras framework and TensorFlow was developed. Independent models were developed for each subtype, and the best performance was achieved for each model after grid search and hyperparameter tuning and balancing of various evaluation metrics, as shown in the table below.

Metrics 1A2 2C9 2C19 2D6 3A4
SE 0.91 0.83 0.83 0.95 0.81
SP 0.90 0.92 0.89 0.83 0.94
ACC 0.91 0.91 0.89 0.86 0.91
AUC 0.97 0.95 0.94 0.96 0.95


Step 1: Provide a string of SMILES format.


Step 1: Upload a file of SMILES format.

Step 2: Insert the verifyCode and press the predict button.


None of the molecule that being uploaded will be retained on the system.


If any collaboration needed, please contact the program instructor Dr. Li: