AndroPred: an artificial intelligence-based model for predicting androgen receptor inhibitors

Area of Science:

• Oncology
• Computational Chemistry
• Drug Discovery

Background:

• The androgen receptor (AR) is crucial in prostate cancer (PCa) development, driving cell proliferation and survival.
• Current AR-inhibiting drugs for PCa are limited due to the difficulty and cost of identifying novel inhibitors.
• Developing new AR inhibitors is essential for advancing prostate cancer treatment.

Purpose of the Study:

• To employ artificial intelligence (AI) algorithms to predict novel androgen receptor (AR) inhibitors for prostate cancer (PCa).
• To accelerate the drug discovery process for AR-targeted therapies.

Main Methods:

• Utilized a dataset of 2242 compounds to train prediction models.
• Applied four machine learning (ML) and deep learning (DL) algorithms.
• Models were trained using molecular descriptors, including 1D, 2D, and molecular fingerprints.

Main Results:

• A deep learning (DL) based prediction model demonstrated superior performance.
• Achieved high accuracies of 92.18% on the training dataset and 93.05% on the test dataset.
• The DNN model showed significant potential for predicting AR inhibitors.

Conclusions:

• Deep learning (DL), particularly the DNN model, is a powerful and effective approach for predicting AR inhibitors.
• This AI-driven strategy can significantly streamline the identification of novel AR inhibitors in prostate cancer drug discovery.
• Further experimental validation is recommended to confirm the predictive accuracy and practical applicability of these models.