QMOD: physically meaningful QSAR.

Area of Science:

• Computational chemistry
• Structural bioinformatics
• Drug discovery

Background:

• Traditional quantitative structure-activity relationship (QSAR) methods for predicting ligand affinity often rely on molecular features unrelated to the actual binding event.
• These methods assume independence and additivity of substituent effects, limiting their accuracy for prospective drug design, especially with diverse chemical scaffolds or significant structural variations.

Purpose of the Study:

• To introduce and evaluate the Surflex-QMOD approach for predicting ligand affinity by constructing physical models of protein binding sites.
• To demonstrate the capability of QMOD in handling complex structure-activity relationships, including non-additive effects.

Main Methods:

• The Surflex-QMOD approach constructs physical models of protein binding sites.
• The method was tested using congeneric CDK2 inhibitors and muscarinic antagonists to assess predictive accuracy.

Main Results:

• Induced binding pockets generated by QMOD can align well with enzyme active sites.
• Model predictivity within a chemical series does not solely depend on the congruence of induced pockets.
• QMOD accurately predicted ligand activity for muscarinic antagonists, even with significant non-additive structure-activity effects.

Conclusions:

• The Surflex-QMOD approach provides a more physically realistic modeling strategy for ligand affinity prediction compared to traditional QSAR.
• This method overcomes limitations of non-physical assumptions in QSAR, enabling more reliable prospective predictions in drug discovery.