Autoencoder-based drug synergy framework for malignant diseases.

Area of Science:

• Computational biology
• Bioinformatics
• Machine learning in drug discovery

Background:

• Drug combinations offer improved efficacy, reduced toxicity, and overcome resistance in treating malignant diseases compared to monotherapy.
• The empirical exploration of potential drug combinations is challenging due to the vast combinatorial space.
• Machine learning and deep learning methods are increasingly employed to identify synergistic drug combinations within large datasets.

Purpose of the Study:

• To propose AESyn, a novel autoencoder-based framework for predicting drug synergy in malignant diseases.
• To utilize a bag-of-words encoding technique for extracting drug-targeted genes and drug features.
• To evaluate the framework's performance using classification and regression metrics and compare it with existing methods.

Main Methods:

• Developed AESyn, an autoencoder-based framework utilizing bag-of-words encoding to represent drug-targeted genes.
• Inputted drug embeddings and drug-targeted genes into autoencoders for feature extraction.
• Trained and validated the framework using screening data from the NCI-ALMANAC and O'Neil datasets.
• Evaluated performance using classification and regression metrics, including accuracy, AUROC, and MAPE.

Main Results:

• The proposed AESyn framework achieved high predictive performance.
• Achieved an accuracy of 95% and an Area Under the Receiver Operating Characteristic curve (AUROC) of 94.2%.
• Demonstrated a Mean Absolute Percentage Error (MAPE) of 7.2, indicating precise regression predictions.

Conclusions:

• The autoencoder-based AESyn framework provides a stable and order-independent method for predicting drug synergy in malignant diseases.
• The framework effectively extracts drug features and predicts synergistic combinations, offering a promising computational approach.
• AESyn demonstrates superior performance compared to existing methods, paving the way for more efficient drug discovery in oncology.