Prediction-driven matched molecular pairs to interpret QSARs and aid the molecular optimization process

Area of Science:

• Computational chemistry and cheminformatics.
• Drug discovery and development.

Background:

• Quantitative Structure-Activity Relationship (QSAR) models offer in silico assessment of chemical properties but lack interpretability.
• Matched Molecular Pairs (MMPs) provide straightforward chemical interpretation of molecular transformations affecting activity.
• Medicinal chemists need practical guidance for molecule synthesis and hit-to-lead optimization.

Purpose of the Study:

• To develop and validate a combined QSAR and MMP approach for enhanced drug design.
• To provide medicinal chemists with interpretable insights into molecular optimization.
• To identify significant molecular transformations impacting biological activities.

Main Methods:

• Prediction-driven MMP analysis integrating QSAR predictions with MMP transformations.
• Application to large chemical datasets for identifying "interesting" molecular changes.
• Development of software tools available online via OCHEM (http://ochem.eu/).

Main Results:

• Demonstrated the utility of prediction-driven MMP analysis as a tool for medicinal chemists.
• Identified numerous "interesting" molecular transformations to drive optimization processes.
• Facilitated the interpretation of complex QSAR models.

Conclusions:

• The prediction-driven MMPs methodology aids in interpreting QSAR models and identifying significant molecular transformations.
• Successfully applied to model aquatic toxicity and CYP3A4 inhibition.
• Facilitates drug design and molecular optimization through interpretable transformations.