Bond Assignment and Utility Tools#

This page covers Enerzyme CLI utilities and Enerzymette helper scripts that support Enerzyme workflows but are not part of core training.

Bond order assignment (enerzyme bond)#

Guess bond orders for enzymatic cluster structures from a PDB file. Compatible with QuantumPDB cluster outputs.

enerzyme bond -p cluster.pdb -m cluster.mol -i cluster.png -t ligands.sdf

Arguments:

  • -p — input PDB (QuantumPDB cluster build output)

  • -m — output MOL file with assigned bonds

  • -i — optional 2D structure image

  • -t — template SDF (e.g. QuantumPDB ligands.sdf)

Requires RDKit. Use the resulting MOL/SDF as reference connectivity for Extracting Fragments by Local Uncertainty or QM input preparation.

Dataset preprocessing only (enerzyme collect)#

Run Datahub preprocessing and splitting without training:

enerzyme collect -c train.yaml -o .

See Preparing a Neural Network Potential Dataset.

Enerzymette: IDPP path interpolation#

Build an initial minimum-energy path between reactant and product for NEB:

enerzymette idpp -r reactant.xyz -p product.xyz -o path.xyz -n 25 -c terachem.inp

  • Uses ASE IDPP interpolation (J. Chem. Phys. 2014, 140, 214106)

  • -c — TeraChem input containing fixed-atom constraints applied during interpolation

  • If reactant/product XYZ contain trajectories, the last frame is used

Use the output multi-frame XYZ as System.structure_file for task: neb, or let Enerzyme interpolate from two endpoints (Running Simulations with a Trained Model). IDPP complements built-in interpolation.method: idpp in NEB configs.

Enerzymette: TeraChem timing#

enerzymette terachem_timing -f terachem.out

Summarizes wall time across SCF iterations and flags incomplete jobs. Useful before merging QC data into training sets (QM Data Annotation).

Enerzymette: ORCA ↔ TeraChem bridge#

enerzymette orca_terachem_request -i orca.extinp.tmp -t terachem_template.inp

Combines ORCA ExtOpt with TeraChem gradients. Full setup is documented in QM Data Annotation.

Enerzymette: flexible scan launcher#

Automated opt → scan → opt loops for reaction-coordinate exploration:

enerzymette enerzyme_scan \
    -r reactant.xyz \
    -o scan_out/ \
    -m model_dir/ \
    -q scan_config.yaml \
    -pp sammt \
    -psc cv_params.yaml \
    -n 25

  • -q — TeraChem input with constraint_freeze / constraint_scan, or a YAML scan config (see Enhanced Sampling and Hybrid Potentials)

  • -pp / -psc — PLUMED CV scan; both flags required together

  • enerzymette update_terachem_scan — refresh bond-scan coordinates in a TeraChem input after geometry update

Enerzymette: NEB launcher#

NNP-driven NEB via ORCA and enerzyme listen:

enerzymette enerzyme_neb \
    -r reactant.xyz \
    -p product.xyz \
    -o neb_out/ \
    -m model_dir/ \
    -q reference.in \
    -c server.yaml \
    -n 25 -b 5000

See Prediction Server and Enerzymette Integration for server setup.

Quick reference#

  • enerzyme bond (Enerzyme) — PDB to MOL connectivity

  • enerzyme collect (Enerzyme) — preprocess and split only

  • enerzymette idpp (Enerzymette) — NEB initial path

  • enerzymette terachem_timing (Enerzymette) — QC job monitoring

  • enerzymette orca_terachem_request (Enerzymette) — ORCA optimizer with TeraChem QC

  • enerzymette enerzyme_scan (Enerzymette) — batch PLUMED or bond-distance scans

  • enerzymette enerzyme_neb (Enerzymette) — NNP-driven NEB via ORCA ExtOpt

  • enerzymette update_terachem_scan (Enerzymette) — refresh TeraChem scan coordinates

  • enerzymette enerzyme_active_learning (Enerzymette) — active-learning campaign launcher

Install Enerzymette with pip install -e . in its repository; it is not bundled with Enerzyme.