CLI tool reference

Each page below documents a single CLI command with common options and examples. For the authoritative, always-up-to-date help, run <command> --help.

calc-v4 calc-v5 prep analyze report sketch series picklists similarity featurize retro schema

Calculators

molprop-calc-v4

Baseline medchem property table from SMILES (CNS MPO, alerts, oral filters, etc.).

molprop-calc-v5

Extended table with solubility/permeability/PK + optional 3D shape descriptors.

Ligand preparation (guide)

Recommended standardization sequences: salts/charges → tautomers → stereo → protomers → 3D.

Tip

If you plan to use 3D descriptors or USR/USRCAT clustering, decide stereochemistry and protonation state up front. The toolkit records those choices in Stereo_* and Protomer_* columns when enabled.

Analysis & reporting

molprop-analyze

Master analyzer: list categories, preview a category, or generate a text report.

molprop-report

Builds a Markdown + HTML report bundle with plots.

molprop-sketch

2D structure depictions from a results CSV, with optional HTML index.

molprop-series

Series analytics: Murcko scaffolds + similarity clustering.

Operational triage

molprop-picklists

Generate decision picklists (CSV and optional HTML) from filters/sorts.

molprop-retro

Offline retrosynthesis wrapper using AiZynthFinder (summary CSV/JSON + HTML route browser).

molprop-similarity

Fingerprint similarity search, pairwise matrices, diversity picking, and simple clustering.

Tip

For MolProp results tables, similarity defaults to the toolkit structure-of-record SMILES and will prefer Calc_Canonical_SMILES when present.

molprop-featurize

Model-ready features: fingerprints + fragments + functional groups + DL exports.

validate schema (utility)

Validate that a results CSV matches the toolkit schema and required categories.

Tip

For large libraries, generate features from Canonical_SMILES (standardized parent) for stability, and keep any protomer/stereo enumeration as a separate dataset variant.

Note

Deep-learning exports include packed-bit fingerprints, aligned SMILES for token models, and concatenated graph tensors for GNN pipelines.