DeepDTA: deep drug-target binding affinity prediction.

Area of Science:

• Computational chemistry
• Bioinformatics
• Drug discovery

Background:

• Drug-target (DT) interaction identification is crucial for drug discovery.
• Current computational methods primarily focus on binary classification of DT interactions, neglecting binding affinity prediction.
• Predicting binding affinity, a continuum of interaction strength, remains a significant challenge.

Purpose of the Study:

• To develop a deep learning model for predicting drug-target binding affinities using only sequence information.
• To explore the efficacy of convolutional neural networks (CNNs) in modeling protein sequences and compound representations for affinity prediction.
• To address the limitations of existing methods that rely on 3D structures or 2D compound features.

Main Methods:

• A deep learning model was developed utilizing convolutional neural networks (CNNs).
• The model processes 1D sequence information from both drug and target molecules.
• High-level representations of drugs and targets were constructed using CNNs.

Main Results:

• The proposed deep learning model effectively predicts drug-target binding affinities using sequence data.
• The CNN-based model achieved the best Concordance Index (CI) performance on benchmark datasets.
• The model outperformed established algorithms like KronRLS and SimBoost.

Conclusions:

• Deep learning models, particularly those employing CNNs on sequence data, offer a powerful approach for drug-target binding affinity prediction.
• This method provides an effective alternative to approaches requiring complex structural information.
• The developed model demonstrates significant potential in advancing drug discovery processes.