Get the name of an organism, given an assembly ID

get_assembly_name_from_ncbi.py

from Bio import Entrez
import xmltodict

def get_name_from_assembly_id(assembly_name):
    handle = Entrez.esearch("assembly", term=assembly_name)
    search_result = xmltodict.parse("".join(handle.readlines()))
    handle.close()
    try:
        assembly_id = search_result["eSearchResult"]["IdList"]["Id"]
    except:
        print("Problem fetching " + assembly_name)
        
        assert False, json.dumps(search_result, indent=4)
    
    # Link to BioSample
    handle = Entrez.elink(db_from="assembly", id=assembly_id, linkname="assembly_biosample")
    link_result = xmltodict.parse("".join(handle.readlines()))
    handle.close()
    try:
        if isinstance(link_result["eLinkResult"]["LinkSet"], list):
            biosample_id = link_result["eLinkResult"]["LinkSet"][0]["LinkSetDb"]["Link"]["Id"]
        else:
            biosample_id = link_result["eLinkResult"]["LinkSet"]["LinkSetDb"]["Link"]["Id"]
    except:
        print("Problem fetching " + assembly_name)
        assert False, json.dumps(link_result, indent=4)
    
    # Fetch from BioSample
    handle = Entrez.efetch(db="biosample", id=biosample_id)
    fetch_result = xmltodict.parse("".join(handle.readlines()))
    handle.close()
    try:
        org_name = fetch_result["BioSampleSet"]["BioSample"]["Description"]["Organism"]["@taxonomy_name"]
    except:
        print("Problem fetching " + assembly_name)
        assert False, json.dumps(fetch_result, indent=4)
    
    
    return "{} ({})".format(org_name, assembly_name)
    
get_name_from_assembly_id("GCF_000332875")