def download_all_tables():
with open("organisms.csv", "r") as f:
for line in f.readlines()[1:]:
organism, taxid = line.strip("\n").split(",")
print("Retrieving %s (taxid %s)" % (organism, taxid))
target = os.path.join("tables", "%s_%s.csv" % (organism, taxid))
download_codons_table(taxid=taxid, target_file=target)
def test_readme_example():
table = pct.get_codons_table("b_subtilis_1423")
assert table["T"]["ACA"] == 0.4
assert table["*"]["TAA"] == 0.61
# LOAD ALL TABLES AT ONCE
codons_tables = pct.get_all_available_codons_tables()
assert codons_tables["c_elegans_6239"]["L"]["CTA"] == 0.09
# GET A TABLE DIRECTLY FROM THE INTERNET
table = pct.download_codons_table(taxid=316407)
assert table["*"]["TGA"] == 0.29
with pytest.raises(RuntimeError):
table = pct.download_codons_table(taxid=000000000) # does not exist
def test_readme_example():
table = pct.get_codons_table("b_subtilis_1423")
assert table['T']['ACA'] == 0.4
assert table['*']['TAA'] == 0.61
# LOAD ALL TABLES AT ONCE
codons_tables = pct.get_all_available_codons_tables()
assert codons_tables['c_elegans_6239']['L']['CTA'] == 0.09
# GET A TABLE DIRECTLY FROM THE INTERNET
table = pct.download_codons_table(taxid=316407)
assert table['*']['UGA'] == 0.29
def codon_tables(taxid, table_path=None):
"""Download the codon use table for the given species and return it as
a dictionary.
Returns:
int: The NCBI taxonomy ID for the supplied species.
Args:
taxid (int): NCBI taxonomy ID for the desrired species.
table_path (str): Defaults to None. Path to a JSON-formatted file representing the
codon usage to consider. If None, the table is fetched from the internet.
Raises:
ValueError: If the NCBI taxonomy ID is not associated with a codon
usage table, raise a ``ValueError`` informing the user and directing
them to the NCBI Taxonomy Browser.
Returns:
dict{str, float}: A dictionary with codons as keys and the frequency
that the codon is used to encode its amino acid as values.
"""
if table_path is None:
try:
taxid = int(taxid)
except ValueError as exc:
taxid = _tax_id_from_species(taxid, exc)
logger.info(
"Downloading host table for NCBI taxonomy ID {}...".format(taxid))
codon_table_by_aa = pct.download_codons_table(taxid)
else:
# load table from disk -- JSON format
with open(table_path, "r") as table:
codon_table_by_aa = json.load(table)
return_dict = {}
for _, codon_dict in codon_table_by_aa.items():
for codon, frequency in codon_dict.items():
return_dict[codon] = frequency
if not return_dict:
raise ValueError(
'"{}" is not a valid host id. '.format(taxid) +
"Supported hosts (Latin and NCBI taxonomy IDs) can be found at " +
_tax_id_url)
return return_dict
def save_codon_table_to_disk(taxid, outfile):
"""Download a codon usage table and save it to disk for future use.
Args:
taxid (str): Name of the species to map.
outfile (str): Full path to the file to write to store the codon
usage information.
"""
try:
taxid = int(taxid)
except ValueError as exc:
taxid = _tax_id_from_species(taxid, exc)
logger.info(
"Downloading host table for NCBI taxonomy ID {}...".format(taxid))
codon_table_by_aa = pct.download_codons_table(taxid)
if not outfile.endswith(".json"):
logger.warning(
'Output file "{}" does not have a JSON file extension.'.format(
outfile))
with open(outfile, "w") as of:
json.dump(codon_table_by_aa, of)
> python codon_usage_retriever.py 316407 e_coli_codon_usage.csv
To retrieve codon tables from all organisms in ``organisms.csv`` at once, use:
> python codon_usage_retriever.py all
"""
import sys
import os
from python_codon_tables import download_codons_table
def download_all_tables():
with open("organisms.csv", "r") as f:
for line in f.readlines()[1:]:
organism, taxid = line.strip("\n").split(",")
print("Retrieving %s (taxid %s)" % (organism, taxid))
target = os.path.join("tables", "%s_%s.csv" % (organism, taxid))
download_codons_table(taxid=taxid, target_file=target)
if __name__ == "__main__":
print(" ".join(sys.argv))
if sys.argv[1] == "all":
download_all_tables()
else:
download_codons_table(sys.argv[1], sys.argv[2])
def test_download_codon_table(tmpdir):
table = pct.download_codons_table(taxid=316407)
assert table['*']['UAG'] == 0.07
target = os.path.join(str(tmpdir), 'test.csv')
table = pct.download_codons_table(taxid=316407, target_file=target)
def test_download_codon_table(tmpdir):
table = pct.download_codons_table(taxid=316407)
assert table["*"]["UAG"] == 0.07
target = os.path.join(str(tmpdir), "test.csv")
table = pct.download_codons_table(taxid=316407, target_file=target)