Prediction of Atropisomerism for Drug-like Molecules

Area of Science:

• Computational chemistry
• Organic chemistry
• Drug discovery

Background:

• Atropisomers are isomers that can be interconverted by rotation about single bonds.
• Accurate classification of atropisomers is crucial for drug discovery and development.
• Existing methods for atropisomer classification can be time-consuming and computationally expensive.

Purpose of the Study:

• To develop a multistep computational workflow for accurate atropisomer classification.
• To achieve a balance between speed and accuracy in computational chemistry for drug discovery.

Main Methods:

• Utilized a multistep computational workflow involving force field (OPLS4) and neural network (QRNN-TB) energy scans.
• Employed transition state searches and density functional theory (DFT) calculations (ωB97X-D3/def2-TZVP(-f)).
• Benchmarked accuracy against high-level computational methods (ωB97M-V/cc-pVTZ and DLPNO-CCSD(T)/def2-TZVPP).

Main Results:

• Developed an automated protocol for assigning organic molecules to three atropisomer classes.
• Achieved a success rate greater than 90% on a test set of 65 molecules.
• Demonstrated excellent correlations when benchmarking predicted rotational barriers.

Conclusions:

• The computational workflow provides an accurate and efficient method for atropisomer classification.
• The developed protocol is suitable for high-throughput screening in drug discovery programs.
• This approach offers a promising balance of speed and accuracy for practical applications.