Drug Db Pubchem

PubChem Database Query

Goal

To programmatically query the PubChem Compound database using the PUG-REST API and retrieve:

• PubChem Compound IDs (CIDs) from names, SMILES, InChI, InChIKey, or molecular formulas,
• computed molecular properties (e.g., molecular weight, XLogP, TPSA, HBD/HBA),
• optional synonyms (names/identifiers),
• optional structure files (SDF), preferring 3D records when available.

This skill is designed for reproducible, rate-limited queries suitable for automation workflows.

Instructions

1. Search by Compound Name

Look up a compound by its common name. Use --name_type complete (default) for exact match or --name_type word for partial matching.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --name "aspirin" \
  --name_type complete \
  --max_results 5 \
  --outdir research/pubchem/aspirin \
  --output aspirin.json

For partial name matching (can be noisier):

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --name "atorvastatin" \
  --name_type word \
  --max_results 10 \
  --outdir research/pubchem/atorvastatin \
  --output atorvastatin_word.json

2. Search by SMILES

SMILES may contain characters reserved by URL syntax; this script uses HTTP POST to avoid common failures.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --smiles "CC(=O)Oc1ccccc1C(=O)O" \
  --max_results 5 \
  --outdir research/pubchem/aspirin_smiles \
  --output aspirin_smiles.json

3. Search by CID

Most unambiguous lookup method.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --cid 2244 \
  --outdir research/pubchem/CID_2244 \
  --output cid_2244.json

4. Search by InChI or InChIKey

InChI uses HTTP POST (like SMILES) to avoid URL syntax issues.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --inchikey "BSYNRYMUTXBXSQ-UHFFFAOYSA-N" \
  --outdir research/pubchem/aspirin_inchikey \
  --output aspirin_inchikey.json

5. Search by Molecular Formula

Uses fastformula for synchronous molecular formula search. Optionally allow additional elements for broader results.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --formula "C9H8O4" \
  --max_results 10 \
  --outdir research/pubchem/C9H8O4 \
  --output formula_results.json

Allow other elements (broader search):

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --formula "C6H12O6" \
  --allow_other_elements \
  --max_results 10 \
  --outdir research/pubchem/C6H12O6_allow_other \
  --output formula_allow_other.json

6. Download SDF Structures (2D/3D)

PubChem 3D records are computationally generated and may be unavailable for some CIDs; the script falls back to 2D by default.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --name "ibuprofen" \
  --download_sdf \
  --sdf_record_type 3d \
  --outdir research/pubchem/ibuprofen \
  --output ibuprofen.json

7. Disable Synonyms

Synonyms require extra API calls; disable them for high-throughput workflows.

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --cid 2244 \
  --no_synonyms \
  --outdir research/pubchem/CID_2244_minimal \
  --output cid_2244_minimal.json

Examples

Caffeine (download 3D SDF if available):

# Env: base-agent
python .agents/skills/drug-db-pubchem/scripts/query_pubchem.py \
  --name "caffeine" \
  --download_sdf \
  --sdf_record_type 3d \
  --outdir research/pubchem/caffeine \
  --output caffeine.json

Constraints

• Rate Limiting: PubChem enforces per-user limits (~5 requests/sec, plus per-minute limits). Exceeding limits triggers HTTP 503 responses. The script rate-limits via a sliding window and retries with exponential backoff. It also adapts to PubChem's dynamic throttling feedback via the X-Throttling-Control header.
• Request Time Limit: PUG-REST is intended for short synchronous requests (server timeouts ~30 seconds). Keep --max_results small for interactive use.
• 3D Structures: PubChem 3D SDF records are computed (not necessarily experimental) and may not exist for all compounds; the script falls back to 2D when needed.
• Synonyms: Synonyms may be numerous and are optional; disable them with --no_synonyms for high-throughput workflows.
• Environment: Requires base-agent conda environment.
• Dependencies: Standard library only (urllib, json, argparse).

---

---

Author: Matthew Cox Contact: GitHub @mcox3406

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: learningmatter-mit
• Source: learningmatter-mit/AtomisticSkills
• License: MIT
• Homepage: https://arxiv.org/abs/2605.24002

Install and usage instructions live in the source repository linked above.