Predicting cyclohexane/water distribution coefficients for the SAMPL5 challenge using MOSCED and the SMD solvation...

Area of Science:

• Physical Chemistry
• Computational Chemistry

Background:

• Accurate prediction of distribution coefficients is crucial for chemical process design and drug discovery.
• The SAMPL5 challenge benchmarks the performance of computational methods for predicting these properties.

Purpose of the Study:

• To evaluate the blind prediction performance of the MOSCED (solubility parameter based method) for cyclohexane/water distribution coefficients.
• To assess the MOSCED method's accuracy and ranking against other computational approaches in the SAMPL5 challenge.

Main Methods:

• Utilized the MOSCED method, parameterized using solvation free energy calculations from electronic structure theory (SMD continuum solvent).
• Approximated distribution coefficients with partition coefficients of neutral species for simplicity.
• Applied the method to a set of 53 compounds for the SAMPL5 challenge.

Main Results:

• Achieved an average unsigned error of [Formula: see text] log units (ranking 15/62).
• Obtained a correlation coefficient (R) of [Formula: see text] (ranking 35/62).
• [Formula: see text] of predictions had the correct sign (ranking 30/62).

Conclusions:

• The MOSCED method provides a viable approach for predicting cyclohexane/water distribution coefficients.
• MOSCED's broad applicability extends to predicting temperature-dependent activity coefficients and phase equilibria.