Data-driven toxicity prediction in drug discovery: Current status and future directions

Area of Science:

• Computational toxicology and cheminformatics.
• Drug discovery and development.
• Data science and machine learning applications in pharmacology.

Background:

• Early toxicity assessment is critical in drug discovery, significantly impacting candidate attrition rates.
• Advancements in information technology have accelerated the development of computational toxicity prediction methods.
• Understanding and mitigating drug toxicity is essential for efficient and safe pharmaceutical development.

Purpose of the Study:

• To provide a comprehensive overview of the current landscape of data-driven toxicity prediction.
• To analyze the features and challenges associated with toxicity prediction.
• To review the evolution of modeling approaches and available tools in the field.

Main Methods:

• Systematic review of existing literature on data-driven toxicity prediction.
• Analysis of research status, challenges, and proposed solutions for toxicity prediction.
• Categorization of methods based on features, modeling approaches, and tools.

Main Results:

• Detailed examination of the characteristics and inherent difficulties in predicting chemical toxicity.
• Tracing the historical development and current trends in computational toxicology modeling.
• Identification and summary of currently available software and platforms for toxicity prediction.

Conclusions:

• Data-driven toxicity prediction is a rapidly evolving field with significant potential to improve drug discovery.
• Addressing existing challenges requires continued innovation in methodologies and tool development.
• Future research should explore novel directions to enhance the accuracy and applicability of predictive models.