SAMPL7 physical property prediction from EC-RISM theory

Area of Science:

• Computational Chemistry
• Physical Organic Chemistry
• Chemical Thermodynamics

Background:

• The embedded cluster reference interaction site model (EC-RISM) combines quantum mechanics with 3D RISM theory for predicting species in solution.
• This method was successfully applied to acidity constants (pK_a) and octanol-water partition coefficients (log P) in previous SAMPL challenges (SAMPL6.1 and SAMPL6.2).

Purpose of the Study:

• To apply the EC-RISM methodology to predict aqueous pK_a and octanol-water log P values for the SAMPL7 physical property challenge.
• To compute distribution coefficients (log D_7.4) using predicted pK_a and log P data.
• To evaluate the performance and identify areas for improvement in the EC-RISM model for these physical properties.

Main Methods:

• Utilized the embedded cluster reference interaction site model (EC-RISM) for quantum-mechanical calculations and 3D RISM theory.
• Applied the methodology to predict pK_a and log P values for the SAMPL7 challenge.
• Calculated log D_7.4 values by combining independent pK_a and log P predictions.

Main Results:

• Macroscopic pK_a predictions showed excellent agreement with experimental data (RMSE = 0.72 pK units).
• Log P predictions exhibited a higher RMSE (1.84), falling short of expectations from previous challenges.
• Log D_7.4 predictions were reasonable (RMSE = 1.69), benefiting from the combination of independent calculations.

Conclusions:

• The EC-RISM model demonstrates strong predictive power for aqueous pK_a values.
• Further refinement of the EC-RISM model is necessary to improve the accuracy and robustness of octanol-water partition coefficient (log P) predictions.
• Continued efforts are needed to align computational model descriptions with experimental conditions for enhanced predictive accuracy in physical property challenges.