Pubchem Database Skill

Source: https://github.com/aipoch/medical-research-skills

When to Use

• You need to search for chemical compounds by name, CID, SMILES, InChI, or molecular formula.
• You want to retrieve physicochemical properties (e.g., molecular weight, LogP, TPSA, H-bond donors/acceptors).
• You need to perform structure-based searches, such as similarity or substructure queries.
• You want to obtain bioactivity data (e.g., assay summaries, target information) for a given compound.
• You are building an automated cheminformatics or drug discovery workflow that requires programmatic access to PubChem.

Key Features

• Flexible compound search by name, CID, SMILES, InChI, or formula.

• Property retrieval via PubChem PUG-REST and PubChemPy (e.g., MW, LogP, Canonical SMILES).

• Structure search:

  • Similarity search
  • Substructure search

• Bioactivity retrieval linked to PubChem BioAssay records.

• Rate-limit aware implementation (respects PubChem’s limit of max 5 requests/sec).

• Python function interface for seamless integration into scientific pipelines.

Dependencies

Install the required Python packages:

uv pip install pubchempy requests

• pubchempy (version: not pinned)
• requests (version: not pinned)

Example Usage

Primary module:

• scripts/pubchem_ops.py

1) Get compound properties

python -c "from scripts.pubchem_ops import get_properties; print(get_properties(query_value='Aspirin', query_type='name'))"

Or in Python:

from scripts.pubchem_ops import get_properties

result = get_properties(query_value="Aspirin", query_type="name")
print(result)

2) Structure search (similarity)

python -c "from scripts.pubchem_ops import structure_search; print(structure_search(query_value='CC(=O)OC1=CC=CC=C1C(=O)O', search_type='similarity'))"

Or in Python:

from scripts.pubchem_ops import structure_search

smiles = "CC(=O)OC1=CC=CC=C1C(=O)O"
result = structure_search(query_value=smiles, search_type="similarity")
print(result)

3) Get bioactivity data

python -c "from scripts.pubchem_ops import get_bioactivity; print(get_bioactivity(cid=2244))"

Or in Python:

from scripts.pubchem_ops import get_bioactivity

result = get_bioactivity(cid=2244)
print(result)

Implementation Details

• Primary script: scripts/pubchem_ops.py

• Data sources / endpoints:

  • Compound & properties: pubchem.ncbi.nlm.nih.gov/rest/pug
  • Bioactivity: PubChem BioAssay endpoints
  • Python wrapper: PubChemPy

• Supported operations:

  • get_properties: retrieve physicochemical properties by name/CID/SMILES/InChI/formula.
  • structure_search: perform similarity or substructure search.
  • get_bioactivity: retrieve assay and bioactivity-related data by CID.

• Input constraints:

  • query_type must match supported types (e.g., name, cid, smiles, inchi, formula).
  • search_type must be similarity or substructure.

• Error handling:

  • Returns structured error or None if compound is not found.
  • Handles PubChem rate limits (≤ 5 requests/sec).

• Troubleshooting considerations:

  • Ensure network access to pubchem.ncbi.nlm.nih.gov.
  • Verify query format (e.g., valid SMILES or InChI) if results are empty.

• Additional reference:

  • API documentation pointers: references/api_reference.md