Rapid Access to Small Molecule Conformational Ensembles in Organic Solvents Enabled by Graph Neural Network-Based...

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Understanding molecular conformations in solvents is crucial for drug discovery and organic synthesis.
• Explicit solvent molecular dynamics (MD) simulations are accurate but computationally expensive.
• Implicit solvent models offer speed but often lack accuracy.

Purpose of the Study:

• To develop a rapid and accurate method for computing small molecule conformational ensembles in diverse organic solvents.
• To overcome the limitations of computational cost and accuracy in existing solvent models.

Main Methods:

• A graph neural network-based implicit solvent (GNNIS) approach was developed.
• The GNNIS model was used to compute conformational ensembles in 39 common organic solvents.
• The accuracy of GNNIS was validated against explicit-solvent simulations and nuclear magnetic resonance (NMR) measurements.

Main Results:

• The GNNIS approach accurately reproduced explicit-solvent simulation results for conformational ensembles.
• Computation time for predicting conformational ensembles was reduced from days to minutes.
• Results were within 1 kBT of experimental values, aiding conformer identification.

Conclusions:

• The GNNIS method provides a computationally efficient and accurate alternative for studying molecular conformations in solution.
• This approach can accelerate drug discovery and organic synthesis by enabling rapid prediction of solvent effects on molecular behavior.