Making priors a priority.

Area of Science:

• Drug Discovery
• Computational Chemistry
• Pharmacology

Background:

• Predictive models for drug properties are rigorously assessed for accuracy.
• The practical utility of these models in decision-making is often overlooked.
• Assessing model utility requires understanding the prior probability distribution of negative outcomes.

Purpose of the Study:

• To illustrate the importance of prior probabilities in evaluating predictive model utility.
• To demonstrate how prior knowledge enhances decision-making in various drug discovery contexts.
• To highlight the impact of prior probabilities on selecting and prioritizing drug candidates.

Main Methods:

• Conceptual analysis and illustration of predictive model utility.
• Exploration of different drug discovery scenarios (selection, prioritization, multi-property balancing).
• Emphasis on the role of prior probabilities of adverse events.

Main Results:

• Prior probabilities significantly influence the practical usefulness of predictive models.
• Understanding priors allows for better compound selection and elimination strategies.
• Incorporating priors improves prioritization for expensive experimental screens.

Conclusions:

• A thorough understanding of prior probabilities is essential for effective drug discovery decision-making.
• Integrating prior knowledge enhances the efficiency of identifying high-quality drug molecules.
• This approach leads to more informed and successful drug development pipelines.