RDKit Conformer Generator

The code is a tiny CLI RDKit wrapper that: (1) builds multiple conformers, (2) aligns them, (3) selects the conformer whose atomic coordinates are closest to the mean (the “median” conformer), and (4) writes that to *.pdb. I use this script to generate structures for high‑quality rendering in PyMol and create publication-ready images. Like here:

---

Requirements

• Python 3.12
• RDKit

Installation / Getting Started

Clone the repo:

git clone https://github.com/markusorsi/confgen.git
cd confgen

Create virtual environment (venv recommended):

python3.12 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

And you're ready to go!

---

Command‑line Usage

python confgen.py -i <SMILES> -o <output_basename> [--nconfs N] [--maxiters M] [--verbose]

Arguments

• -i, --input (required): SMILES string to generate conformers from.
• -o, --output (required): Output basename (script appends .pdb).
• --nconfs (default 10): Number of conformers to generate.
• --maxiters (default 500): Maximum MMFF optimization iterations per conformer.
• --verbose: Print details about conformer minimization.

Example

# Aspirin
python confgen.py -i "CC(=O)OC1=CC=CC=C1C(=O)O" -o aspirin --nconfs 50 --verbose

Output:

Mean conformer saved to outputs/aspirin.pdb

---

How it works

1. Parse & hydrogenate the input SMILES (Chem.MolFromSmiles + AddHs).
2. Generate multiple conformers with EmbedMultipleConfs (reproducible seed, chirality enforced, knowledge‑based torsions enabled).
3. Optimize each conformer using MMFF (MMFFOptimizeMolecule).
4. Align conformers for analysis/visualization.
5. Compute the median conformer: find the conformer with minimal squared distance to the mean atomic coordinates across all conformers.
6. Export that conformer to PDB (Chem.MolToPDBFile).

---

Rendering tips

• PyMOL:

  load aspirin.pdb
  hide everything
  show sticks
  hide (hydro) # Optional hide hydrogens
  util.cbaw # Colors by atom type 
  ray 2400,1800
  png aspirin.png

---

The resulting image will resemble the one shown at the beginning of the README.

Python API

If you’d like to call the functions directly because you need a quick conformer generator in one of your project pipelines:

from conformers import ConformersFromSmiles, ComputeMedianConformer
from rdkit import Chem

mol = ConformersFromSmiles('CC(=O)OC1=CC=CC=C1C(=O)O', nConfs=50, maxIters=1000)
median = ComputeMedianConformer(mol)
Chem.MolToPDBFile(median, 'aspirin.pdb')

---

Notes & Caveats

• The “median” here is the conformer closest to the mean coordinate set; it is a representative conformer, not necessarily the global minimum.
• Reproducibility is controlled via randomSeed=42 in EmbedMultipleConfs.

Tip: If you plan to add an aligned multi‑model PDB export (for comparing conformers), your helper ConfsToAlignedMolsList already has most of the pieces.

---

License

MIT License

---

Acknowledgments

The phenomenal open‑source cheminformatics toolkit RDKit.

---