Spectrum of deep learning algorithms in drug discovery

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Artificial intelligence in drug discovery

Background:

• Deep learning (DL) algorithms, a subset of machine learning, model complex relationships.
• Advances in understanding disease mechanisms coincide with DL applications in drug discovery.
• High-throughput methods necessitate automation, large data management, and data fusion.

Purpose of the Study:

• To review the impact of DL in drug discovery and design workflows.
• To present the types of DL algorithms used in this field.
• To discuss the pros and cons of DL algorithms and future research directions.

Main Methods:

• Literature review of DL applications in drug discovery.
• Analysis of DL algorithms employed in various stages of drug development.
• Discussion of challenges and future trends in AI-driven drug discovery.

Main Results:

• DL aids in data management, reaction optimization, molecule construction, and structure/interaction prediction.
• DL accelerates exploration of drug mechanisms and drug repositioning.
• DL integration streamlines preclinical and clinical studies.

Conclusions:

• Deep learning significantly impacts and accelerates the drug discovery and design process.
• Various DL algorithms offer distinct advantages and disadvantages for specific applications.
• Future research should focus on further integrating DL for enhanced automation and data interpretation in drug development.