The current limits in virtual screening and property prediction.

Area of Science:

• Computational chemistry
• Pharmacology
• Drug discovery

Background:

• Optimizing drug candidates involves addressing metabolic, toxicological, and off-target effects beyond target binding affinity.
• Experimental synthesis and testing of drug candidates are resource-intensive.
• Virtual screening is a crucial computational tool in preclinical drug development.

Purpose of the Study:

• To highlight the indispensable role of virtual screening in preclinical drug development.
• To discuss the applications of virtual screening in predicting molecular properties and identifying potential issues.
• To outline challenges and limitations associated with current computational models in virtual screening.

Main Methods:

• Application of virtual screening for predicting molecular properties like solubility and metabolic liability.
• Utilizing virtual screening to predict binding to antitargets, such as the hERG channel.
• Exploring emerging aspects of virtual screening, including prediction of binding sites and drugable targets.

Main Results:

• Virtual screening effectively predicts crucial molecular properties relevant to drug development.
• The approach aids in identifying potential liabilities, including metabolic issues and off-target interactions (e.g., hERG channel binding).
• Emerging applications show promise in predicting drug targets and binding sites.

Conclusions:

• Virtual screening significantly reduces the experimental burden in preclinical drug development.
• Computational models, including machine learning, face challenges in prediction accuracy and overfitting.
• Continued development of virtual screening methods is essential for efficient drug discovery and optimization.