Command-line tools¶

Installing CReM adds the following console commands to your PATH.

Command overview¶

Command	Purpose	Guide
cremdb_create	Build a v1 database from SMILES (recommended)	Build (v1)
cremdb_convert	Convert a v0 database to v1	Convert
cremdb_add_prop	Add property columns to a database	Properties
cremdb_get_set_names	List the fragment sets in a database	Fragment sets
cremdb_merge	Merge shard databases into one	Build (v1)
fragmentation	v0 pipeline — fragment molecules	Build (v0)
frag_to_env	v0 pipeline — standardize context/core	Build (v0)
env_to_db	v0 pipeline — import into a v0 database	Build (v0)
crem_create_frag_db.sh	v0 pipeline — all-in-one shell wrapper	Build (v0)
guacamol_test	Run the GuacaMol benchmark	Benchmarks

---

cremdb_get_set_names¶

List the fragment-set columns in each radius table.

cremdb_get_set_names -i fragments.db

Option	Description
-i, --input	Path to the database (required)

---

guacamol_test¶

Run the GuacaMol goal-directed benchmark with a CReM-based generator. Requires the optional guacamol package (and pandas, joblib, numpy).

guacamol_test --smiles_file train.smi --db_fname fragments.db --output_dir results/

Key options: --smiles_file, --db_fname, --selection_size (10), --radius (3), --replacements (1000), --min_size (0), --max_size (10), --min_inc (-7), --max_inc (7), --generations (1000), --ncpu (1), --seed (42), --suite (v2), --output_dir. Results are written as goal_directed_results.json plus per-task .smi files. See Benchmarks.