| name | conformation_generation_dft_input |
| description | Generating 3D molecular conformations and preparing input files for quantum chemistry DFT calculations. |
Conformation Generation & DFT Input Preparation
Use this skill when initializing the computational chemistry pipeline from a 2D representation (SMILES) to obtain 3D coordinates for quantum mechanical calculations (DFT) or Machine Learning Interatomic Potentials (MLIP) input.
Conceptual Framework
3D Conformation Generation
Molecules exist as dynamic ensembles of 3D conformations (conformers) in physical space. For property prediction and force field training, we require realistic 3D geometries.
- ETKDG Method: RDKit's Experimental-Torsion Knowledge Distance Geometry (ETKDG) method is used to generate initial conformer coordinates based on experimental crystal structure databases.
- Force Field Relaxation: Generated conformers are optimized using a fast empirical force field (e.g., MMFF94) to eliminate steric clashes and produce physically realistic starting structures.
- Quantum Chemistry Refinement (DFT): The lowest-energy conformer(s) are selected and exported to higher-fidelity Density Functional Theory (DFT) calculations for ground-truth properties (energies and forces).
Recommended Workflow
- SMILES to 3D Mol: Convert SMILES to an RDKit Mol object and add hydrogens.
- Embed Conformers: Use
AllChem.EmbedMultipleConfs with ETKDG parameters.
- Minimize Energy: Use
AllChem.MMFFOptimizeMolecule to relax the geometries.
- Export XYZ: Extract coordinates of the lowest energy conformer.
- Generate Input Deck: Create an ORCA or Gaussian input file.
Execution
Run the script congenerate_dft.py to generate coordinates and an ORCA input deck for a SMILES string:
python scripts/congenerate_dft.py --smiles "CCO" --num_conformers 5
