What problem does it solve?
This Skill generates structurally related variants from a starting molecule to accelerate lead optimization, scaffold hopping, and analog exploration while controlling similarity and scaffold preservation constraints.
Core Features & Use Cases
- Single-molecule sampling: Create new molecules from one seed SMILES with configurable priors and similarity thresholds.
- Batch sampling: Apply the same generation procedure to multiple seed molecules to produce per-seed derivative libraries.
- Flexible priors and filtering: Support for similarity, medium_similarity, high_similarity, scaffold, scaffold_generic, and mmp priors with Lipinski and filter presets.
- Rich outputs for downstream analysis: Produces validated SMILES lists and a CSV report containing similarity, MW, LogP, HBD, HBA, TPSA, and Lipinski violations for programmatic verification.
- Operational guidance: Includes recommended n values, prior selection guidance, scaffold-preservation cautions, and mandatory generation count verification to avoid data fabrication.
Quick Start
Generate 30 scaffold-preserving analogs for the provided SMILES using the scaffold_generic prior and min_similarity 0.5.
