def test_get_genome(tmpdir):
collection = GenomeCollection(data_dir=str(tmpdir))
taxid = PHAGE_TAXID
path = collection.datafile_path(taxid, data_type="genomic_fasta")
assert not os.path.exists(path)
genome_path = collection.get_taxid_genome_data_path(taxid)
assert path == genome_path
file_size = os.stat(genome_path).st_size
assert 200_000 > file_size > 150_000
def test_taxid_with_nonunique_genome(tmpdir):
taxid = 10710 # LAMBDA PHAGE - has 2 associated assemblies on NCBI
collection = GenomeCollection(data_dir=str(tmpdir))
with pytest.raises(OSError) as excinfo:
path = collection.get_taxid_infos(taxid)
assert "You will need to download" in str(excinfo.value)
collection.download_taxid_genome_infos_from_ncbi(taxid, assembly_id="#1")
path = collection.datafile_path(taxid, data_type="genomic_fasta")
assert not os.path.exists(path)
collection.get_taxid_genome_data_path(taxid)
assert os.path.exists(path)
def test_get_blast_database(tmpdir):
collection = GenomeCollection(data_dir=str(tmpdir))
taxid = PHAGE_TAXID
path = collection.datafile_path(taxid, data_type="blast_nucl")
assert not os.path.exists(path + ".nsq")
# Test nucleotide database
blast_db_path = collection.get_taxid_blastdb_path(taxid, db_type="nucl")
assert path == blast_db_path
file_size = os.stat(blast_db_path + ".nsq").st_size
assert 50_000 > file_size > 30_000
# Test protein database
blast_db_path = collection.get_taxid_blastdb_path(taxid, db_type="prot")
file_size = os.stat(blast_db_path + ".psq").st_size
assert 60_000 > file_size > 40_000