Bio Conformer Generation

Version Compatibility

Reference examples tested with: RDKit 2024.09+, xtb 6.7+, CREST 3.0+, OpenMM 8.1+ for follow-up MD.

Before using code patterns, verify installed versions match. If versions differ:

• Python: pip show <package> then help(module.function) to check signatures
• CLI: xtb --version; crest --version

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Conformer Generation

Generate 3D conformer ensembles for molecules from 2D structures. The choice of method depends on molecule size, flexibility, and downstream use: ETKDG (Riniker & Landrum 2015) and its ETKDGv3 macrocycle update (Wang et al. 2020) are modern defaults for drug-like molecules, MMFF94/UFF provide fast energy minimization, and CREST + GFN2-xTB provide higher-cost semi-empirical sampling. A single conformer may be insufficient when the downstream result is conformation-sensitive; determine ensemble size by convergence of the downstream descriptor, alignment, or docking result.

For docking pose validation, see chemoinformatics/pose-validation. For free-energy methods (which require ensemble sampling), see chemoinformatics/free-energy-calculations.

Conformer Method Taxonomy

Decision: Start drug-like organic molecules with ETKDGv3 and a parameter-checked MMFF94/MMFF94s optimization. Escalate difficult macrocycles, peptides, or highly flexible molecules to a validated CREST workflow when downstream convergence is inadequate. Benchmark ML generators on the intended chemistry before using them at scale.

Decision Tree by Scenario

ETKDGv3 (Modern Default)

ETKDGv3 (Wang et al. 2020), building on the original ETKDG method (Riniker & Landrum 2015), incorporates experimental torsion preferences and updated macrocycle handling into distance geometry.

Goal: Generate an ensemble of 3D conformers from a SMILES with the modern default embedding algorithm.

Approach: Add explicit hydrogens, configure ETKDGv3 parameters (random seed, maximum embedding iterations, random coordinates), and embed multiple conformers via EmbedMultipleConfs.

from rdkit import Chem
from rdkit.Chem import AllChem

def gen_conformers(smiles, n_conf=20, seed=42):
    mol = Chem.MolFromSmiles(smiles)
    mol = Chem.AddHs(mol)
    params = AllChem.ETKDGv3()
    params.randomSeed = seed
    params.useRandomCoords = True
    params.maxIterations = 1000
    ids = AllChem.EmbedMultipleConfs(mol, numConfs=n_conf, params=params)
    return mol, list(ids)

useRandomCoords=True can improve convergence for macrocycles and highly flexible molecules. On an EmbedParameters object the documented limit is maxIterations; maxAttempts is only present in legacy positional overloads.

Force-Field Optimization

After embedding, minimize each conformer to a local minimum.

Goal: Reduce strain in each embedded conformer to a stable local minimum and record the resulting energies.

Approach: Build MMFF94s force-field parameters, minimize each conformer in place, and collect energies; fall back to UFF when MMFF94 cannot parameterize the molecule.

def optimize_conformers(mol, conf_ids, force_field='mmff94'):
    results = []
    if force_field == 'mmff94':
        mmff_props = AllChem.MMFFGetMoleculeProperties(mol, mmffVariant='MMFF94s')
        if mmff_props is not None:
            for cid in conf_ids:
                ff = AllChem.MMFFGetMoleculeForceField(mol, mmff_props, confId=cid)
                if ff is None:
                    results.append({'conf_id': cid, 'status': 'force_field_failed'})
                    continue
                status = ff.Minimize(maxIts=1000)
                results.append({'conf_id': cid, 'energy': ff.CalcEnergy(),
                                'converged': status == 0, 'force_field': 'MMFF94s'})
            return results
        force_field = 'uff'
    if force_field == 'uff':
        if not AllChem.UFFHasAllMoleculeParams(mol):
            raise ValueError('Neither MMFF94s nor UFF covers this molecule')
        for cid in conf_ids:
            ff = AllChem.UFFGetMoleculeForceField(mol, confId=cid)
            if ff is None:
                results.append({'conf_id': cid, 'status': 'force_field_failed'})
                continue
            status = ff.Minimize(maxIts=1000)
            results.append({'conf_id': cid, 'energy': ff.CalcEnergy(),
                            'converged': status == 0, 'force_field': 'UFF'})
        return results
    raise ValueError(f'Unsupported force field: {force_field}')

MMFF94 vs MMFF94s: MMFF94s is the static-structure variant, with modified out-of-plane and torsional terms intended to preserve planarity in selected functional groups. It is useful for geometry optimization, but it is not a universally preferred replacement for MMFF94; record which variant was used and validate it for the downstream task.

UFF (Universal Force Field): UFF has different and often broader parameter coverage than MMFF94, but RDKit does not guarantee coverage for every molecule and generic UFF does not validate metal coordination chemistry. Call UFFHasAllMoleculeParams before use and use a chemistry-appropriate method for metal complexes.

RMSD Pruning

Remove near-duplicate conformers within a chosen RMSD cutoff to keep the ensemble diverse:

import numpy as np

def prune_conformers_rmsd(mol, conf_ids, rmsd_cutoff=0.5):
    n = len(conf_ids)
    keep = []
    for i, cid in enumerate(conf_ids):
        is_unique = True
        for kept_cid in keep:
            rmsd = AllChem.GetBestRMS(mol, mol, cid, kept_cid)
            if rmsd < rmsd_cutoff:
                is_unique = False
                break
        if is_unique:
            keep.append(cid)
    return keep

Typical RMSD cutoff (Source / Rationale):

Energy Window Filtering

Remove conformers above a project-justified energy cutoff only when the energy model and downstream purpose support that choice. Bound conformers can be strained relative to an isolated-molecule minimum.

def filter_by_energy(mol, conf_ids, energies, window_kcal=10.0):
    min_e = min(energies)
    keep = []
    for cid, e in zip(conf_ids, energies):
        if e - min_e <= window_kcal:
            keep.append(cid)
    return keep

Treat any numerical energy window as a starting parameter. Calibrate it against conformer recovery or downstream metric convergence and record the energy method, solvent treatment, protonation state, and temperature assumptions.

Macrocycle Handling

Macrocycles (>=12 atom rings) have distinct conformational issues: ETKDGv3 default knowledge base under-samples macrocycle torsions. Use macrocycle-specific torsion preferences:

from rdkit.Chem import AllChem

def macrocycle_conformers(smiles, n_conf=200, seed=42):
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError(f'Invalid SMILES: {smiles!r}')
    mol = Chem.AddHs(mol)
    params = AllChem.ETKDGv3()
    params.randomSeed = seed
    params.useRandomCoords = True
    params.useMacrocycleTorsions = True
    params.useSmallRingTorsions = True
    params.maxIterations = 5000
    ids = list(AllChem.EmbedMultipleConfs(mol, numConfs=n_conf, params=params))
    if not ids:
        raise RuntimeError(f'No macrocycle conformers embedded for {smiles!r}')
    return mol, ids

For difficult macrocycles, CREST + GFN2-xTB is a useful higher-cost option; validate coverage against experimental or downstream evidence rather than treating one method as universally definitive.

CREST + GFN2-xTB for High-Quality Sampling

CREST (Pracht et al. 2024) performs iterative meta-dynamics + GFN2-xTB optimization for conformer sampling.

Goal: Sample high-quality conformer ensembles for macrocycles, peptides, or molecules where ETKDGv3 + MMFF94 is inadequate.

Approach: Start from an RDKit-generated MMFF94-relaxed conformer, write to XYZ, and run CREST with GFN2-xTB driver to perform iterative meta-dynamics + reoptimization.

xtb mol.xyz --opt extreme
crest xtbopt.xyz --gfn2 --T 12 --ewin 6

--gfn2: use GFN2-xTB; validate its coverage and energy ordering for the chemistry of interest. --gfn-ff: use GFN-FF; it can reduce computational cost, but benchmark its coverage and energy ordering for the intended molecules. -ewin 6: 6 kcal/mol energy window above global min. -T 12: use 12 CPU threads.

Output: crest_conformers.xyz with sampled ensemble.

Workflow: Start from RDKit ETKDGv3 + MMFF94 (cheap initial structure) -> save as XYZ -> CREST refinement.

from rdkit import Chem
from rdkit.Chem import AllChem
import subprocess
from pathlib import Path

def crest_workflow(smiles, out_dir='crest_out'):
    mol = Chem.MolFromSmiles(smiles)
    if mol is None:
        raise ValueError(f'Invalid SMILES: {smiles!r}')
    mol = Chem.AddHs(mol)
    params = AllChem.ETKDGv3()
    params.useRandomCoords = True
    if AllChem.EmbedMolecule(mol, params) == -1:
        raise RuntimeError(f'Initial 3D embedding failed for {smiles!r}')
    if not AllChem.MMFFHasAllMoleculeParams(mol):
        raise ValueError(f'MMFF parameters are unavailable for {smiles!r}')
    status = AllChem.MMFFOptimizeMolecule(
        mol, mmffVariant='MMFF94s', maxIters=1000
    )
    if status != 0:
        raise RuntimeError(f'Initial MMFF optimization did not converge (status {status})')

    out_dir = Path(out_dir).resolve()
    out_dir.mkdir(parents=True, exist_ok=True)
    input_path = out_dir / 'input.xyz'
    input_path.write_text(Chem.MolToXYZBlock(mol))
    # Because cwd is out_dir, pass the coordinate filename rather than out_dir/input.xyz.
    subprocess.run(['crest', input_path.name, '--gfn2', '-T', '12'],
                   cwd=out_dir, check=True)
    output_path = out_dir / 'crest_conformers.xyz'
    if not output_path.exists():
        raise FileNotFoundError(f'CREST did not produce {output_path}')
    return output_path

Boltzmann Averaging of Properties

For ensemble descriptors (3D shape, dipole moment, polar surface area in 3D), Boltzmann-weight by energy:

import numpy as np

def boltzmann_weights(energies, T=300.0):
    energies = np.array(energies)
    kt = 0.001987 * T  # kcal/mol at 300K
    rel = energies - energies.min()
    w = np.exp(-rel / kt)
    return w / w.sum()

def boltzmann_average(values, energies, T=300.0):
    w = boltzmann_weights(energies, T)
    return float(np.sum(np.array(values) * w))

Boltzmann populations require energies that approximate the relevant thermodynamic state, including conformer degeneracy and environmental effects where important. Raw gas-phase MMFF/UFF minima are exploratory surrogates, not generally validated populations.