Computational methods for early predictive safety assessment from biological and chemical data

Area of Science:

• Drug Discovery and Development
• Computational Toxicology
• Pharmacology

Background:

• Early predictive safety assessment (PSA) aims to identify compound liabilities to prevent progression in the drug development pipeline.
• Undesirable compounds can hinder pharmaceutical research and delay the delivery of novel therapeutics.
• Computational methods are increasingly vital for understanding safety data, generating hypotheses, and supplementing experimental procedures.

Purpose of the Study:

• To survey computational and data-driven methods for early predictive safety assessment.
• To review approaches for predicting compound liabilities from structure, adverse event data, and toxicology studies.

Main Methods:

• Review of methods for predicting liabilities of individual compounds based on their structures.
• Analysis of adverse event data and safety profiling panels.
• Examination of toxicology data mining from animal experiments.

Main Results:

• Computational methods offer valuable tools for safety pharmacology and can reduce reliance on traditional experimental procedures.
• Diverse methods exist for assessing compound safety at different scales and complexities.
• Mining toxicology data from animal studies provides insights into potential liabilities.

Conclusions:

• A major challenge for data-driven safety assessment is the lack of integrated datasets and standardized ontologies.
• Improved data sharing and the use of standard ontologies are essential for effective safety evaluations.
• Informed decisions and enhanced hit assessments in screening campaigns can lead to the early termination of undesirable compounds.