Database lookup

Database Lookup

Part of Agent Skills™ by googleadsagent.ai™

Description

Database Lookup provides unified programmatic access to 78+ scientific and public databases spanning chemistry (PubChem, ChEMBL), biology (UniProt, COSMIC, Ensembl), clinical (ClinicalTrials.gov, FDA), economics (FRED, World Bank), and intellectual property (USPTO, EPO). The agent constructs API queries, handles pagination, normalizes responses, and caches results for reproducible research workflows.

Scientific research increasingly depends on integrating data from multiple heterogeneous databases. A drug discovery project might query ChEMBL for bioactivity data, UniProt for target protein information, PubChem for compound properties, ClinicalTrials.gov for related clinical studies, and FRED for healthcare spending trends—all for a single research question. This skill abstracts the API differences into a unified query interface.

Each database connector handles authentication, rate limiting, response parsing, and error recovery. Results are normalized into consistent schemas (DataFrames with typed columns) regardless of the source API's format (REST JSON, XML, CSV, SPARQL). Caching prevents redundant API calls and enables offline analysis of previously retrieved data.

Use When

• Retrieving compound data from PubChem or ChEMBL
• Querying protein sequences or annotations from UniProt
• Searching clinical trials on ClinicalTrials.gov
• Fetching economic indicators from FRED or World Bank
• Looking up patent information from USPTO
• Integrating data across multiple scientific databases

How It Works

graph TD
    A[Research Query] --> B[Query Router]
    B --> C{Database Selection}
    C -->|Chemistry| D[PubChem / ChEMBL / DrugBank]
    C -->|Biology| E[UniProt / Ensembl / COSMIC]
    C -->|Clinical| F[ClinicalTrials.gov / FDA / OMIM]
    C -->|Economics| G[FRED / World Bank / BLS]
    C -->|Patents| H[USPTO / EPO / WIPO]
    D --> I[API Request + Rate Limiting]
    E --> I
    F --> I
    G --> I
    H --> I
    I --> J[Response Normalization]
    J --> K[Cache Layer]
    K --> L[Unified DataFrame Output]

The query router identifies the appropriate database based on the query type and entity. All responses pass through normalization to produce consistent DataFrames with standardized column names and types.

Implementation

import requests
import pandas as pd
from functools import lru_cache
from time import sleep

class DatabaseClient:
    BASE_URLS = {
        "pubchem": "https://pubchem.ncbi.nlm.nih.gov/rest/pug",
        "chembl": "https://www.ebi.ac.uk/chembl/api/data",
        "uniprot": "https://rest.uniprot.org/uniprotkb",
        "clinicaltrials": "https://clinicaltrials.gov/api/v2/studies",
        "fred": "https://api.stlouisfed.org/fred/series/observations",
    }

    def __init__(self, cache_dir: str = ".db_cache"):
        self.session = requests.Session()
        self.session.headers["User-Agent"] = "AgentSkills/1.0 (research)"

    def pubchem_compound(self, name: str) -> dict:
        url = f"{self.BASE_URLS['pubchem']}/compound/name/{name}/JSON"
        resp = self._get(url)
        props = resp["PC_Compounds"][0]["props"]
        return {
            "cid": resp["PC_Compounds"][0]["id"]["id"]["cid"],
            "name": name,
            "properties": {p["urn"]["label"]: p["value"] for p in props},
        }

    def chembl_target(self, uniprot_id: str) -> pd.DataFrame:
        url = f"{self.BASE_URLS['chembl']}/target.json"
        resp = self._get(url, params={
            "target_components__accession": uniprot_id,
            "limit": 100,
        })
        return pd.json_normalize(resp["targets"])

    def uniprot_search(self, query: str, limit: int = 25) -> pd.DataFrame:
        url = f"{self.BASE_URLS['uniprot']}/search"
        resp = self._get(url, params={
            "query": query,
            "format": "json",
            "size": limit,
            "fields": "accession,id,protein_name,organism_name,length,sequence",
        })
        return pd.json_normalize(resp["results"])

    def clinical_trials(self, condition: str, status: str = "RECRUITING") -> pd.DataFrame:
        url = self.BASE_URLS["clinicaltrials"]
        resp = self._get(url, params={
            "query.cond": condition,
            "filter.overallStatus": status,
            "pageSize": 50,
        })
        return pd.json_normalize(resp["studies"])

    def fred_series(self, series_id: str, api_key: str) -> pd.DataFrame:
        url = self.BASE_URLS["fred"]
        resp = self._get(url, params={
            "series_id": series_id,
            "api_key": api_key,
            "file_type": "json",
        })
        df = pd.DataFrame(resp["observations"])
        df["value"] = pd.to_numeric(df["value"], errors="coerce")
        df["date"] = pd.to_datetime(df["date"])
        return df

    def _get(self, url: str, params: dict = None) -> dict:
        sleep(0.25)
        resp = self.session.get(url, params=params, timeout=30)
        resp.raise_for_status()
        return resp.json()

Best Practices

• Respect rate limits: 5 req/s for PubChem, 1 req/s for ChEMBL, 3 req/s for UniProt
• Cache all API responses locally to enable offline analysis and reduce server load
• Normalize identifiers (CID, ChEMBL ID, UniProt accession) before cross-database joins
• Handle pagination for large result sets—never assume all results fit in one response
• Log every API query for reproducibility, including timestamp and response hash
• Set a User-Agent header identifying your tool and contact information

Platform Compatibility

Platform	Support	Notes
Cursor	Full	Python + HTTP client
VS Code	Full	REST client integration
Windsurf	Full	API query support
Claude Code	Full	Database query generation
Cline	Full	API integration
aider	Partial	Code-level support

Related Skills

• Bioinformatics
• Cheminformatics
• Data Analysis
• Knowledge Base RAG

Keywords

database-lookup pubchem chembl uniprot clinical-trials fred scientific-databases api-integration data-retrieval

---

© 2026 googleadsagent.ai™ | Agent Skills™ | MIT License