openbabel - SKILL.md Agent Skill

name: openbabel description: Use when converting molecular file formats, generating 3D coordinates, searching conformers, computing descriptors/fingerprints, or filtering chemical libraries with OpenBabel. Covers both pybel Python API and obabel command-line tool.

OpenBabel — Chemical Format Conversion & Manipulation

OpenBabel 3.1.1. Two interfaces: pybel (Python API, high-level) and obabel (CLI, batch processing). Supports 146 formats, MMFF94/UFF/GAFF force fields.

When to Use This Skill

Converting between molecular formats: SMILES ↔ SDF ↔ MOL2 ↔ PDB ↔ InChI ↔ CIF ↔ XYZ ↔ 100+ others
Generating 3D coordinates from SMILES (quick alternative to RDKit ETKDGv3)
Conformer searching with force field scoring
Protonation state at given pH
Computing molecular descriptors (LogP, TPSA, MR) and fingerprints
SMARTS substructure filtering of large libraries
Batch library processing (split, deduplicate, filter)
Converting formats unsupported by RDKit (CIF, VASP POSCAR, XYZ, etc.)

Quick Start

from openbabel import pybel

# Read SMILES → generate 3D → write SDF
mol = pybel.readstring('smi', 'CC(=O)Oc1ccccc1C(=O)O')   # aspirin
mol.make3D(forcefield='mmff94', steps=500)
mol.write('sdf', 'aspirin.sdf', overwrite=True)

# Read SDF → SMILES
for mol in pybel.readfile('sdf', 'library.sdf'):
    print(mol.write('can').strip())   # canonical SMILES

Router — What to Read

Task	Reference
pybel Python API: read, write, 3D, descriptors, fingerprints, SMARTS	`references/pybel-python.md`
obabel CLI: conversion, --gen3d, --conformer, filtering, pH, split	`references/obabel-cli.md`
Format codes, fingerprint types, descriptors, Tanimoto	`references/formats-fingerprints.md`

Two Interfaces

	pybel	obabel CLI
Use case	Scripted workflows, per-molecule logic	Batch conversion, library filtering
Import	`from openbabel import pybel`	`subprocess` or shell
Speed	Moderate	Fast (C++ core)
Flexibility	High (per-atom access)	Moderate (flags)

Installation

conda install -c conda-forge openbabel   # recommended (includes C++ libs)
pip install openbabel                     # Linux/macOS only

# Verify
python -c "from openbabel import pybel; print(pybel.readstring('smi','C').molwt)"
obabel --version

Key Global Variables

from openbabel import pybel

pybel.informats    # dict: {'sdf': 'MDL MOL format', 'smi': 'SMILES format', ...}
pybel.outformats   # dict of writable formats
pybel.fps          # list of fingerprint types: ['FP2', 'FP3', 'FP4', 'MACCS']
pybel.descs        # list of descriptor names
pybel.forcefields  # list of available force fields

Relation to RDKit

Task	Prefer
Drug-like 3D (ETKDGv3)	RDKit
Non-organic / unusual atoms	OpenBabel (UFF)
Format not in RDKit (CIF, XYZ, etc.)	OpenBabel
SMARTS filtering (speed)	OpenBabel CLI
Fingerprints (ECFP)	RDKit
Fingerprints (FP2/FP3/MACCS)	OpenBabel

Related Skills

rdkit — complementary: ETKDGv3 conformers, ECFP fingerprints, reactions
ase — ASE reads XYZ/CIF; OpenBabel converts to those formats
scientific-skills:datamol — fast preprocessing, also wraps RDKit