genesis/api_clients/bioportal_api.py

# genesis/api_clients/bioportal_api.py
import requests
from typing import List, Dict, Optional

BIOPORTAL_BASE_URL = "https://data.bioontology.org"
BIOPORTAL_API_KEY = None  # Load from secrets in Hugging Face or local .env

def search_term(term: str, ontologies: Optional[List[str]] = None, max_results: int = 20) -> List[Dict]:
    """
    Search BioPortal for a term across one or more ontologies.
    If no ontologies specified, searches all available ones.
    """
    headers = {"Authorization": f"apikey token={BIOPORTAL_API_KEY}"}
    params = {
        "q": term,
        "pagesize": max_results
    }
    if ontologies:
        params["ontologies"] = ",".join(ontologies)

    r = requests.get(f"{BIOPORTAL_BASE_URL}/search", headers=headers, params=params)
    r.raise_for_status()
    data = r.json()

    results = []
    for result in data.get("collection", []):
        results.append({
            "pref_label": result.get("prefLabel"),
            "definition": result.get("definition", ["No definition found"])[0],
            "synonyms": result.get("synonym", []),
            "ontology": result.get("links", {}).get("ontology"),
            "iri": result.get("@id"),
            "score": result.get("score", 0)
        })
    return results


def get_concept_details(ontology_acronym: str, concept_id: str) -> Dict:
    """
    Fetch detailed metadata for a specific ontology concept.
    """
    headers = {"Authorization": f"apikey token={BIOPORTAL_API_KEY}"}
    r = requests.get(f"{BIOPORTAL_BASE_URL}/ontologies/{ontology_acronym}/classes/{concept_id}",
                     headers=headers)
    r.raise_for_status()
    data = r.json()

    return {
        "pref_label": data.get("prefLabel"),
        "definition": data.get("definition", ["No definition found"])[0],
        "synonyms": data.get("synonym", []),
        "properties": data.get("properties", {}),
        "relations": data.get("links", {})
    }


def map_to_ontologies(term: str, ontologies: List[str]) -> Dict[str, List[str]]:
    """
    Map a given term to equivalent terms in multiple ontologies.
    Returns a dictionary mapping ontology → list of equivalent labels.
    """
    results = search_term(term, ontologies=ontologies)
    mapping = {}
    for res in results:
        onto_name = res.get("ontology", "Unknown")
        if onto_name not in mapping:
            mapping[onto_name] = []
        mapping[onto_name].append(res.get("pref_label"))
    return mapping