scripts/uniprot.py

import requests, sys
import xml.dom.minidom

def get_protein_location(accession=None, gene=None, protein=None, organism=None):
    requestURL = 'https://www.ebi.ac.uk/proteins/api/proteins?offset=0&size=-1'

    if (accession is not None):
        requestURL = f'{requestURL}&accession={accession}'
    else:
        if (gene is not None):
            requestURL = f'{requestURL}&gene={gene}'
        elif (protein is not None):
            requestURL = f'{requestURL}&protein={protein}'
        else:
            raise ValueError('Either accession, gene, or protein must be specified in the search parameters')
        if organism is not None:
            requestURL = f'{requestURL}&organism={organism}'

    r = requests.get(requestURL, headers={"Accept": "application/xml"})

    if not r.ok:
        r.raise_for_status()
        sys.exit()

    xml_doc = xml.dom.minidom.parseString(r.text)

    packages = xml_doc.getElementsByTagName('subcellularLocation')

    locations = []
    for package in packages:
        locations.append(package.getElementsByTagName('location')[0].childNodes[0].data)

    return list(set(locations))

#def is_transcription_factor(accession=None, gene=None, protein=None, organism=None):