Virtual screening: hope, hype, and the fine line in between.

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• Virtual screening (VS) has rapidly advanced, leading to increased publications in drug discovery.
• A significant gap exists between computational predictions and experimental validation, resulting in unverified hits.
• Machine learning (ML) has enhanced VS integration into drug discovery pipelines.

Purpose of the Study:

• To examine the current VS landscape, including practices, challenges, and limitations.
• To identify strategies for mitigating challenges in VS workflows.
• To explore approaches for optimizing VS hits by addressing pharmacodynamic and pharmacokinetic issues.

Main Methods:

• Review of current VS practices and literature.
• Analysis of case studies and common pitfalls in VS.
• Exploration of strategies for integrating biological and in silico data.
• Examination of ML applications in VS.

Main Results:

• Integrated strategies combining biological and in silico data show higher success rates.
• ML adoption has improved VS integration but lacks standardized evaluation criteria.
• Addressing pharmacodynamic and pharmacokinetic issues is crucial for optimizing VS hits.

Conclusions:

• Standardized evaluation criteria are needed to objectively assess VS studies.
• Effective strategies can mitigate challenges and improve the success rate of VS in drug discovery.
• Optimizing VS hits requires addressing both computational and experimental validation aspects.