Predicting Subtype Selectivity for Adenosine Receptor Ligands with Three-Dimensional Biologically Relevant Spectrum...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Adenosine receptors (ARs) are crucial therapeutic targets for various diseases, including Parkinson's disease, diabetes, pain, stroke, and cancers.
• Accurate prediction of subtype selectivity for ARs is essential for drug development and understanding therapeutic mechanisms.

Purpose of the Study:

• To introduce and validate a novel molecular descriptor, the three-dimensional biologically relevant spectrum (BRS-3D), for predicting adenosine receptor subtype selectivity.
• To compare the performance of BRS-3D with existing 3D descriptors in selectivity prediction models.

Main Methods:

• Development of pairwise regression and discrimination models using support vector machine methods.
• Introduction of BRS-3D, a shape similarity profile, as a molecular descriptor encoding multiple conformations.
• Validation of model robustness and stability through resampling and Y-randomization techniques.

Main Results:

• BRS-3D demonstrated strong performance in predicting AR subtype selectivity, with regression models achieving an average r² of 0.664 on test sets.
• The A2B vs A3 selectivity model showed excellent performance (q²=0.769 training, r²=0.766 test).
• Discrimination models achieved high average accuracy (ACC=0.912) and Matthews correlation coefficient (MCC=0.792) on test sets, outperforming previous methods.

Conclusions:

• BRS-3D is an effective molecular descriptor for predicting adenosine receptor subtype selectivity.
• The developed models show significant potential for guiding the design of selective AR modulators.
• BRS-3D offers a valuable tool for both therapeutic targeting and mechanistic studies of adenosine receptors.