Alterlab Datamol

---
name: alterlab-datamol
description: Wraps RDKit in a high-level, pandas-friendly datamol interface with sensible defaults for everyday drug discovery — SMILES/SDF loading into DataFrames, molecule standardization, descriptors, fingerprints, Butina clustering, 3D conformer generation, scaffold analysis, and parallel batch processing, returning native rdkit.Chem.Mol objects. Use when running standard cheminformatics pipelines on molecule tables with minimal boilerplate; for low-level control, custom sanitization, or specialized algorithms prefer alterlab-rdkit. Part of the AlterLab Academic Skills suite.
license: Apache-2.0
allowed-tools: Read Write Edit Bash(python:*) Bash(uv:*)
compatibility: "Self-contained — runs under `uv run python` with the skill's Python package installed; no API key or account required."
metadata:
    skill-author: AlterLab
    version: "1.0.0"
---

# Datamol Cheminformatics Skill

## Overview

Datamol is a Python library that provides a lightweight, Pythonic abstraction layer over RDKit for molecular cheminformatics. Simplify complex molecular operations with sensible defaults, efficient parallelization, and modern I/O capabilities. All molecular objects are native `rdkit.Chem.Mol` instances, ensuring full compatibility with the RDKit ecosystem.

**Key capabilities**:
- Molecular format conversion (SMILES, SELFIES, InChI)
- Structure standardization and sanitization
- Molecular descriptors and fingerprints
- 3D conformer generation and analysis
- Clustering and diversity selection
- Scaffold and fragment analysis
- Chemical reaction application
- Visualization and alignment
- Batch processing with parallelization
- Cloud storage support via fsspec

## Installation and Setup

Guide users to install datamol:

```bash
uv pip install datamol
```

Examples here are verified against **datamol 0.12.x** (pulls in RDKit automatically). The descriptor key names below are stable in this line; pin if you depend on them: `uv pip install 'datamol>=0.12,<0.13'`.

**Import convention**:
```python
import datamol as dm
```

## Core Workflows

Each subsection below shows the primary call pattern. Full API signatures, parameters, and secondary examples live in the per-module reference files cited under each; complete multi-step pipelines live in `references/workflow_recipes.md`.

### 1. Basic Molecule Handling

```python
import datamol as dm

# Parse SMILES (returns None on failure)
mol = dm.to_mol("CCO")                        # Ethanol
mols = [dm.to_mol(smi) for smi in ["CCO", "c1ccccc1", "CC(=O)O"]]
if dm.to_mol("invalid_smiles") is None:
    print("Failed to parse SMILES")

# Export to common formats (canonical + isomeric by default)
smiles   = dm.to_smiles(mol)                  # keeps stereochemistry
flat     = dm.to_smiles(mol, isomeric=False)  # drops stereochemistry
inchi    = dm.to_inchi(mol)
inchikey = dm.to_inchikey(mol)
selfies  = dm.to_selfies(mol)

# Standardize user-provided molecules (recommended for datasets)
mol = dm.sanitize_mol(mol)
mol = dm.standardize_mol(mol, disconnect_metals=True, normalize=True, reionize=True)
clean_smiles = dm.standardize_smiles(smiles)
```

Full conversion, sanitization, and standardization API: see `references/core_api.md`.

### 2. Reading and Writing Molecular Files

```python
# Read (open_df auto-detects .sdf/.csv/.xlsx/.parquet/.json)
df = dm.read_sdf("compounds.sdf", mol_column='mol')
df = dm.read_csv("data.csv", smiles_column="SMILES", mol_column="mol")
df = dm.open_df("file.sdf")

# Write
dm.to_sdf(mols, "output.sdf")               # or dm.to_sdf(df, "output.sdf", mol_column="mol")
dm.to_smi(mols, "output.smi")
dm.to_xlsx(df, "output.xlsx", mol_columns=["mol"])   # renders molecule images in cells

# Remote paths work everywhere via fsspec (S3, GCS, HTTP)
df = dm.read_sdf("s3://bucket/compounds.sdf")
dm.to_sdf(mols, "s3://bucket/output.sdf")
```

Full reader/writer signatures (`read_smi`, `read_excel`, `read_mol2file`, `read_pdbfile`, `save_df`, shared parameters): see `references/io_module.md`.

### 3. Molecular Descriptors and Properties

```python
# Single molecule -> ~22 keys. Note datamol's naming (NOT rdkit's):
desc = dm.descriptors.compute_many_descriptors(mol)
#   {'mw': 46.04, 'clogp': -0.0, 'n_lipinski_hbd': 1, 'n_lipinski_hba': 1,
#    'tpsa': 20.23, 'n_rotatable_bonds': 0, 'qed': ..., 'fsp3': ..., 'sas': ..., ...}
# Gotcha: logP is 'clogp'; donors/acceptors are 'n_lipinski_hbd'/'n_lipinski_hba'.
# There is no 'logp', 'hbd', 'hba', or 'n_aromatic_atoms' key in this dict.

# Batch (parallel) -> DataFrame with the same keys
desc_df = dm.descriptors.batch_compute_many_descriptors(mols, n_jobs=-1, progress=True)

# Standalone descriptors not in the dict above
dm.descriptors.n_aromatic_atoms(mol)
dm.descriptors.n_stereo_centers(mol)
dm.descriptors.n_rigid_bonds(mol)

# Drug-likeness filter (Lipinski's Rule of Five) with datamol's exact key names
def is_druglike(mol):
    d = dm.descriptors.compute_many_descriptors(mol)
    return (d['mw'] <= 500 and d['clogp'] <= 5 and
            d['n_lipinski_hbd'] <= 5 and d['n_lipinski_hba'] <= 10)

druglike_mols = [m for m in mols if is_druglike(m)]
```

Full descriptor catalog, RDKit descriptor access, and ADME examples: see `references/descriptors_viz.md`.

### 4. Molecular Fingerprints and Similarity

```python
# Fingerprints (ECFP/Morgan is the default)
fp       = dm.to_fp(mol, fp_type='ecfp', radius=2, n_bits=2048)
fp_maccs = dm.to_fp(mol, fp_type='maccs')
# Also available: 'topological', 'atompair', 'fcfp'

# Similarity as Tanimoto distance (distance = 1 - similarity; lower = more similar)
distance_matrix = dm.pdist(mols, n_jobs=-1)                       # within one set
distances       = dm.cdist(query_mols, library_mols, n_jobs=-1)  # between two sets
from scipy.spatial.distance import squareform
dist_matrix = squareform(dm.pdist(mols))                         # square form
```

Fingerprint types and `pdist` / `cdist` details: see `references/core_api.md`.

### 5. Clustering and Diversity Selection

```python
# Butina clustering (cutoff = Tanimoto distance; each cluster is a list of indices)
clusters = dm.cluster_mols(mols, cutoff=0.2, n_jobs=-1)
for i, cluster in enumerate(clusters):
    cluster_mols = [mols[idx] for idx in cluster]

# Diversity / representative selection
diverse   = dm.pick_diverse(mols, npick=100)
centroids = dm.pick_centroids(mols, npick=50)
```

**Scale note**: Butina builds a full distance matrix — fine for ~1,000 molecules, not 10,000+. Clustering parameters: see `references/core_api.md`.

### 6. Scaffold Analysis

```python
# Bemis-Murcko scaffold (core ring systems + linkers)
scaffold = dm.to_scaffold_murcko(mol)
scaffold_smiles = dm.to_smiles(scaffold)
```

Scaffold frequency counting, scaffold-to-molecule grouping, and scaffold-based train/test splitting for ML: see `references/workflow_recipes.md`. `fuzzy_scaffolding` and more: see `references/fragments_scaffolds.md`.

### 7. Molecular Fragmentation

```python
# BRICS (16 bond types) and RECAP (11 bond types) both return SMILES with
# attachment points like '[1*]CCN'
frags_brics = dm.fragment.brics(mol)
frags_recap = dm.fragment.recap(mol)
```

Cross-library fragment frequency analysis and fragment-overlap scoring recipes: see `references/workflow_recipes.md`. MMPA fragmentation and a method comparison table: see `references/fragments_scaffolds.md`.

### 8. 3D Conformer Generation

```python
# Generate 3D conformers (ETKDGv3 recommended; UFF minimization on by default)
mol_3d = dm.conformers.generate(mol, n_confs=50, rms_cutoff=0.5,
                                minimize_energy=True, method='ETKDGv3')
mol_3d.GetNumConformers()
conf = mol_3d.GetConformer(0)
positions = conf.GetPositions()          # Nx3 array of atom coordinates

# Cluster conformers by RMSD and take representatives
clusters  = dm.conformers.cluster(mol_3d, rms_cutoff=1.0, centroids=False)
centroids = dm.conformers.return_centroids(mol_3d, clusters)

# Solvent accessible surface area
sasa_values = dm.conformers.sasa(mol_3d, n_jobs=-1)
sasa = mol_3d.GetConformer(0).GetDoubleProp('rdkit_free_sasa')
```

Embedding methods, RMSD matrices, and low-level coordinate manipulation: see `references/conformers_module.md`.

### 9. Visualization

```python
# Grid image (PNG by default; use_svg=True for publications)
dm.viz.to_image(mols[:20], legends=[dm.to_smiles(m) for m in mols[:20]],
                n_cols=5, mol_size=(300, 300))
dm.viz.to_image(mols, outfile="molecules.png")
dm.viz.to_image(mols, outfile="molecules.svg", use_svg=True)

# Align by MCS for SAR series; highlight atoms/bonds; render conformers
dm.viz.to_image(similar_mols, align=True, legends=activity_labels, n_cols=4)
dm.viz.to_image(mol, highlight_atom=[0, 1, 2, 3], highlight_bond=[0, 1, 2])
dm.viz.conformers(mol_3d, n_confs=10, align_conf=True, n_cols=3)
```

Full `to_image` / `conformers` / `circle_grid` parameters and best practices: see `references/descriptors_viz.md`.

### 10. Chemical Reactions

```python
from rdkit.Chem import rdChemReactions

# Build a reaction from SMARTS, then apply it to a reactant tuple
rxn = rdChemReactions.ReactionFromSmarts('[C:1](=[O:2])[OH:3]>>[C:1](=[O:2])[Cl:3]')
product = dm.reactions.apply_reaction(rxn, (dm.to_mol("CC(=O)O"),), sanitize=True)
product_smiles = dm.to_smiles(product)
```

Batch reaction application, common reaction templates (amide, Suzuki, esterification), and the toy `datamol.data` datasets: see `references/reactions_data.md`.

## Parallelization

Datamol includes built-in parallelization for many operations. Use `n_jobs` parameter:
- `n_jobs=1`: Sequential (no parallelization)
- `n_jobs=-1`: Use all available CPU cores
- `n_jobs=4`: Use 4 cores

**Functions supporting parallelization**:
- `dm.read_sdf(..., n_jobs=-1)`
- `dm.descriptors.batch_compute_many_descriptors(..., n_jobs=-1)`
- `dm.cluster_mols(..., n_jobs=-1)`
- `dm.pdist(..., n_jobs=-1)`
- `dm.conformers.sasa(..., n_jobs=-1)`

**Progress bars**: Many batch operations support `progress=True` parameter.

## Common Workflows and Patterns

Full copy-ready worked pipelines — data loading → filtering → analysis, Structure-Activity Relationship (SAR) analysis, and virtual screening — plus machine-learning feature generation and robust error-handling wrappers, have moved out of this file to keep it lean. See `references/workflow_recipes.md`.

## Reference Documentation

For detailed API documentation, consult these reference files:

- **`references/core_api.md`**: Core namespace functions (conversions, standardization, fingerprints, clustering)
- **`references/io_module.md`**: File I/O operations (read/write SDF, CSV, Excel, remote files)
- **`references/conformers_module.md`**: 3D conformer generation, clustering, SASA calculations
- **`references/descriptors_viz.md`**: Molecular descriptors and visualization functions
- **`references/fragments_scaffolds.md`**: Scaffold extraction, BRICS/RECAP fragmentation
- **`references/reactions_data.md`**: Chemical reactions and toy datasets
- **`references/workflow_recipes.md`**: End-to-end pipelines, SAR/screening recipes, ML integration, error handling

## Best Practices

1. **Always standardize molecules** from external sources:
   ```python
   mol = dm.standardize_mol(mol, disconnect_metals=True, normalize=True, reionize=True)
   ```

2. **Check for None values** after molecule parsing:
   ```python
   mol = dm.to_mol(smiles)
   if mol is None:
       # Handle invalid SMILES
   ```

3. **Use parallel processing** for large datasets:
   ```python
   result = dm.operation(..., n_jobs=-1, progress=True)
   ```

4. **Leverage fsspec** for cloud storage:
   ```python
   df = dm.read_sdf("s3://bucket/compounds.sdf")
   ```

5. **Use appropriate fingerprints** for similarity:
   - ECFP (Morgan): General purpose, structural similarity
   - MACCS: Fast, smaller feature space
   - Atom pairs: Considers atom pairs and distances

6. **Consider scale limitations**:
   - Butina clustering: ~1,000 molecules (full distance matrix)
   - For larger datasets: Use diversity selection or hierarchical methods

7. **Scaffold splitting for ML**: Ensure proper train/test separation by scaffold

8. **Align molecules** when visualizing SAR series

## Troubleshooting

**Issue**: Molecule parsing fails
- **Solution**: Use `dm.standardize_smiles()` first or try `dm.fix_mol()`

**Issue**: Memory errors with clustering
- **Solution**: Use `dm.pick_diverse()` instead of full clustering for large sets

**Issue**: Slow conformer generation
- **Solution**: Reduce `n_confs` or increase `rms_cutoff` to generate fewer conformers

**Issue**: Remote file access fails
- **Solution**: Ensure fsspec and appropriate cloud provider libraries are installed (s3fs, gcsfs, etc.)

## Additional Resources

- **Datamol Documentation**: https://docs.datamol.io/
- **RDKit Documentation**: https://www.rdkit.org/docs/
- **GitHub Repository**: https://github.com/datamol-io/datamol

Part of the AlterLab Academic Skills suite.