Source code for bioregistry.external.uniprot

# -*- coding: utf-8 -*-

"""Download and parse the UniProt Cross-ref database."""

import json
import logging
from typing import Mapping

import requests

from bioregistry.constants import EXTERNAL, URI_FORMAT_KEY

__all__ = [

logger = logging.getLogger(__name__)

#: Download URL for the UniProt registry
URL = "*"
DIRECTORY.mkdir(exist_ok=True, parents=True)
RAW_PATH = DIRECTORY / "raw.json"
PROCESSED_PATH = DIRECTORY / "processed.json"

#: resources with these UniProt prefixes don't exist anymore
skip_prefixes = {
    "UniPathway",  # doesn't exist anymore
    "BRENDA",  # has bad format string contains EC, UniProt, and taxon
    "eggNOG",  # not sure what this does
    "PlantReactome",  # incomprehensible URLs
    "Reactome",  # incomprehensible URLs

[docs]def get_uniprot(force_download: bool = True) -> Mapping[str, Mapping[str, str]]: """Get the UniProt registry.""" if PROCESSED_PATH.is_file() and not force_download: with as file: return json.load(file) RAW_PATH.write_text( json.dumps(requests.get(URL).json(), indent=2, sort_keys=True, ensure_ascii=False) ) rv = {} for record in json.loads(RAW_PATH.read_text())["results"]: processed_record = _process_record(record) if processed_record is None: continue prefix = processed_record.pop("prefix") if prefix in skip_prefixes: continue rv[prefix] = processed_record with"w") as file: json.dump(rv, file, indent=2, sort_keys=True) return rv
def _process_record(record): rv = { "prefix": record.pop("id"), "name": record.pop("name"), "abbreviation": record.pop("abbrev"), "homepage": record.pop("server"), "category": record.pop("category"), } doi = record.pop("doiId", None) pubmed = record.pop("pubMedId", None) publication = {} if doi: publication["doi"] = doi.lower() if pubmed: publication["pubmed"] = pubmed if publication: rv["publications"] = [publication] del record["linkType"] del record["statistics"] rv = {k: v for k, v in rv.items() if k and v} value = record.pop("dbUrl") if "%s" in value and "%u" in value: logger.debug(f"has both formats: {value}") return None else: value = value.replace("%s", "$1").replace("%u", "$1") if "$1" in value: rv[URI_FORMAT_KEY] = value else: logger.debug("no annotation in %s", rv["prefix"]) if record: logger.debug("forgot something: %s", record) return rv if __name__ == "__main__": get_uniprot(force_download=False)