Finding the rules for successful drug optimisation.

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Development

Background:

• Drug discovery involves optimizing multiple parameters to identify successful compounds.
• Project-specific criteria for compound success are often subjective.
• Historical data analysis can inform objective decision-making in drug discovery.

Purpose of the Study:

• To introduce computational rule induction for objective analysis of drug discovery data.
• To identify interpretable, multiparameter rules for successful compound prioritization.
• To apply rule induction for optimizing drug-like properties and reducing toxicity risks.

Main Methods:

• Utilized computational 'rule induction' techniques.
• Analyzed historical drug discovery data to identify success-predictive rules.
• Applied methods to two case studies: drug-like properties and target inhibition data.

Main Results:

• Demonstrated the ability of rule induction to derive objective, interpretable rules from complex datasets.
• Identified critical data points for effective compound prioritization.
• Showcased rule induction's utility in defining rules for drug-like properties and predicting/reducing toxicity.

Conclusions:

• Computational rule induction offers an objective approach to multiparameter optimization in drug discovery.
• This methodology enhances the ability to prioritize compounds with a higher likelihood of success.
• Rule induction can effectively guide decisions to mitigate risks, such as compound toxicity.