Skip to content

Quick Start

This guide walks you through running moTSart on a single reaction using the built-in env=test configuration.

Input Format

moTSart reads CSV files with no header row and two columns:

  1. rxn_id
  2. rxn_smiles

Example:

0,[CH2:1]=[CH:2][CH:3]=[CH2:4].[CH2:5]=[CH2:6]>>[CH2:5]1[CH2:6][CH2:4][CH:3]=[CH:2][CH2:1]1

Test data is provided in data/test_rxns.csv.

Running Individual Steps

Step 1: Complex Finder

python -m motsart.complex_finder.complex_finder env=test env.rxn_num=0

Tip

For fast runs, keep env=test and explicit test configs: afir_cfg=test optim_cfg=test.

Step 2: Path Guessers

Run path guessers in this order:

  1. RMSD-PP to generate initial TS guesses.
  2. RacerTS to refine those RMSD-PP guesses.
python -m motsart.path_guessers.rmsd_pp.rmsd_pp_reaction_path_guesser env=test env.rxn_num=0

python -m motsart.path_guessers.ts_conf_sampler env=test env.rxn_num=0

Step 3: Validator

Validate TS guesses with xTB:

python -m motsart.validator.base_validator env=test validator_cfg=test validator=xtb env.rxn_num=0

Optional: Reproduction Workflow

For model-generated TS evaluation and paper-style comparison workflow, see Paper Reproduction Workflow.

Running the Full Pipeline

bash complex_and_ts_search_local.sh

complex_and_ts_search_local.sh is a local template script. Adjust RXN_NUM, env, and config names for your run.

Parallel Execution

Use Hydra joblib launcher:

python -m motsart.complex_finder.complex_finder -m \
    hydra/launcher=joblib \
    hydra.launcher.n_jobs=2 \
    env=test \
    "env.rxn_num=range(0,2)"

Results

Results are saved under the configured results directory (for env=test, default is results_test/):

results_test/
└── R{rxn_id}/
    ├── r/                       # Reactant complexes
    ├── p/                       # Product complexes
    ├── ts/{method}/ts_to_validate/
    └── validation/{method}/