Database Lookup

# Database Lookup

Part of [Agent Skills™](https://github.com/itallstartedwithaidea/agent-skills) by [googleadsagent.ai™](https://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

```mermaid
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

```python
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__ac