| name | molclaw-denovo-sampling |
| description | Generate entirely new drug-like molecules from scratch without any starting molecule, using REINVENT4's de novo prior. |
| license | MIT license |
| metadata | {"skill-author":"PJLab"} |
De Novo Molecule Generation
When to Use This Tool
Use reinvent_denovo_sampling when you need to generate molecules without any starting structure — pure chemical space exploration. This is useful for building initial compound libraries, generating diverse scaffolds for screening, or creating random drug-like molecules for testing pipelines.
Do NOT use when: you have a starting molecule and want analogs (use reinvent_mol2mol_sampling); you want targeted optimization toward specific properties (use reinvent_similarity_optimization); you need to design linkers, R-groups, or peptides (use the corresponding specialized tools).
Tool Description
Generate new molecules de novo (from scratch, no input molecule required).
Args:
n (int): Number of molecules to sample. Actual valid output is typically 50-80% of n.
lipinski (bool): Whether to apply Lipinski's Rule of Five filtering, default True
filter_preset (str): Filter preset controlling chemical quality. Options:
- 'none': No filtering — raw model output
- 'minimal': Remove only chemically unreasonable structures
- 'default': Standard chemical filters (reactive groups, toxic alerts)
- 'strict': Stringent filters
- 'druglike': Drug-likeness oriented filtering (recommended for drug discovery)
- 'all': Apply all available filters
Default: 'druglike'
Return:
status (str): 'success' or 'error'
msg (str): Descriptive message
save_smiles_file (str): Path to the saved CSV file with generated molecules and computed properties
output_smiles_list (List[str]): List of generated SMILES strings
Recommended Parameters
| Purpose | n | filter_preset | lipinski |
|---|
| Quick test / pipeline validation | 10-20 | 'default' | True |
| Drug-like library building | 100-500 | 'druglike' | True |
| Maximum diversity exploration | 200-1000 | 'minimal' | False |
| Fragment-like library | 50-200 | 'strict' | True |
Usage Example
response = await client.session.call_tool(
"reinvent_denovo_sampling",
arguments={
"n": 100,
"lipinski": True,
"filter_preset": "druglike"
}
)
result = client.parse_result(response)
output_smiles_list = result["output_smiles_list"]
Note on Downstream Use
De novo generated molecules are typically used as starting points for further workflows: property filtering (Skill 3), docking screening (Skill 2), or iterative optimization (Skill 5). Since these molecules have no target bias, expect low hit rates in docking — this is normal for de novo generation.
⚠ Mandatory Generation Count Verification (L3 Principle 11)
After calling this tool, IMMEDIATELY verify the actual number of molecules generated.
actual_count = len(result["output_smiles_list"])
If actual count < 70% of requested: Consider retrying with a less restrictive filter_preset or increased n.
