Deep learning tools for advancing drug discovery and development.

Area of Science:

• Computational chemistry and bioinformatics
• Drug discovery and development
• Artificial intelligence in medicine

Background:

• Traditional drug discovery is time-consuming and expensive, relying heavily on manual laboratory work.
• Advancements in computational techniques and multi-omics data have improved efficiency.
• The integration of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) offers further rationalization of the drug discovery process.

Purpose of the Study:

• To review the application of ML and DL in the drug discovery and development pipeline.
• To introduce ML and DL concepts, learning methods, and training models.
• To summarize existing DL tools and their specific applications in drug discovery.

Main Methods:

• Review of existing literature on ML/DL applications in drug discovery.
• Categorization of DL tools based on their function (e.g., target identification, virtual screening, ADMET prediction).
• Discussion of specific DL models and their roles in various stages of drug development.

Main Results:

• ML/DL approaches leverage big biological data to identify therapeutic molecules more efficiently.
• Numerous DL tools are available for tasks like drug-target interaction prediction (e.g., DeepCPI, DeepDTA), protein structure prediction, and de novo design.
• DL models are applied across the pipeline, including virtual screening, ADMET prediction, and clinical trial design.

Conclusions:

• ML and DL significantly accelerate and optimize the drug discovery and development process.
• The review highlights successful ML/DL models and discusses current challenges and future prospects.
• This review serves as a guide for researchers seeking DL tools for their drug discovery projects.