def invoke_check_fasta_gff():
"""
Parse command-line options then invoke
:py:func:`riboviz.check_fasta_gff.check_fasta_gff`.
"""
print(provenance.write_provenance_to_str(__file__))
options = parse_command_line_options()
fasta = options.fasta
gff = options.gff
check_fasta_gff.check_fasta_gff(fasta, gff)
def test_check_fasta_gff_no_such_gff_file(tmpdir):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff`
with FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and
a non-existent GFF file raises an exception.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
"""
issues_file = tmpdir.join("issues.tsv")
with pytest.raises(FileNotFoundError):
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE, "nosuch.gff",
issues_file)
def test_check_fasta_gff(tmpdir):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff` with
FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and GFF file
(:py:const:`TEST_GFF_CHECK_FILE`) and validate the TSV file
output.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
"""
issues_file = tmpdir.join("issues.tsv")
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE, TEST_GFF_CHECK_FILE,
issues_file)
check_fasta_gff_issues_csv(TEST_CHECK_ISSUES, issues_file)
def test_check_fasta_gff_empty_gff_file(tmpdir):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff`
with FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and
an empty GFF file raises an exception.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
"""
# Cast to str to avoid "'LocalPath' object is not subscriptable"
# error from gff library.
gff_file = str(tmpdir.join("gff.gff"))
open(gff_file, 'a').close()
issues_file = tmpdir.join("issues.tsv")
with pytest.raises(ValueError):
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE, gff_file,
issues_file)
def test_check_fasta_gff_start_codons(tmpdir):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff` with
FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and GFF file
(:py:const:`TEST_GFF_CHECK_FILE`) and custom
``start_codons`` and validate the TSV file output.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
"""
issues_file = tmpdir.join("issues.tsv")
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE,
TEST_GFF_CHECK_FILE,
issues_file,
start_codons=TEST_START_CODONS)
# TEST_START_CODONS includes all start codons in test data so
# expect no NO_START_CODON issues in results.
test_check_issues = [(s, f, i, d) for (s, f, i, d) in TEST_CHECK_ISSUES
if i != check_fasta_gff.NO_START_CODON]
check_fasta_gff_issues_csv(test_check_issues, issues_file)
def test_check_fasta_gff_use_feature_name_true(tmpdir):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff` with
FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and GFF file
(:py:const:`TEST_GFF_CHECK_FILE`) and ``use_feature_name=True``
and validate the TSV file output.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
"""
issues_file = tmpdir.join("issues.tsv")
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE,
TEST_GFF_CHECK_FILE,
issues_file,
use_feature_name=True)
# Create expected results when use_feature_name=True.
test_check_issues = TEST_CHECK_ISSUES.copy()
test_check_issues.remove(TEST_NO_ATG_START_ID_NAME_ATTR_ISSUE)
test_check_issues.append((TEST_NO_ATG_START_ID_NAME_ATTR_ISSUE[0],
TEST_NO_ATG_START_ID_NAME_ATTR_NAME,
TEST_NO_ATG_START_ID_NAME_ATTR_ISSUE[2],
TEST_NO_ATG_START_ID_NAME_ATTR_ISSUE[3]))
check_fasta_gff_issues_csv(test_check_issues, issues_file)
def test_check_fasta_gff_feature_format(tmpdir):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff` with
FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and GFF file
(:py:const:`TEST_GFF_CHECK_FILE`) and custom
``cds_feature_format`` and validate the TSV file output.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
"""
issues_file = tmpdir.join("issues.tsv")
cds_feature_format = "{}-Custom"
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE,
TEST_GFF_CHECK_FILE,
issues_file,
feature_format=cds_feature_format)
# Create expected results for custom cds_feature_format.
test_check_issues = TEST_CHECK_ISSUES.copy()
test_check_issues.remove(TEST_NO_ID_NAME_ATTR_ISSUE)
test_check_issues.append(
(TEST_NO_ID_NAME_ATTR_ISSUE[0],
cds_feature_format.format(TEST_NO_ID_NAME_ATTR_PREFIX),
TEST_NO_ID_NAME_ATTR_ISSUE[2], TEST_NO_ID_NAME_ATTR_ISSUE[3]))
check_fasta_gff_issues_csv(test_check_issues, issues_file)
def test_check_fasta_gff_stdout(tmpdir, capsys):
"""
Test :py:func:`riboviz.check_fasta_gff.check_fasta_gff` with
FASTA file (:py:const:`TEST_FASTA_CHECK_FILE`) and GFF file
(:py:const:`TEST_GFF_CHECK_FILE`) and validate the standard
output.
:param tmpdir: Temporary directory (pytest built-in fixture)
:type tmpdir: py._path.local.LocalPath
:param capsys: Capture standard output (pytest built-in fixture)
:type capsys: _pytest.capture.CaptureFixture
"""
issues_file = tmpdir.join("issues.tsv")
check_fasta_gff.check_fasta_gff(TEST_FASTA_CHECK_FILE,
TEST_GFF_CHECK_FILE,
issues_file,
is_verbose=True)
lines = capsys.readouterr()[0].split("\n")
expected_config = [
"Configuration:", "{}\t{}".format(check_fasta_gff.FASTA_FILE,
TEST_FASTA_CHECK_FILE),
"{}\t{}".format(check_fasta_gff.GFF_FILE, TEST_GFF_CHECK_FILE),
"{}\t{}".format(check_fasta_gff.START_CODONS, [START_CODON]), ""
]
assert lines[0:len(expected_config)] == expected_config,\
"Unexpected Configuration header"
lines = lines[len(expected_config):]
expected_metadata = [
"Metadata:", "{}\t{}".format(check_fasta_gff.NUM_SEQUENCES,
TEST_NUM_SEQUENCES),
"{}\t{}".format(check_fasta_gff.NUM_FEATURES, TEST_NUM_FEATURES),
"{}\t{}".format(check_fasta_gff.NUM_CDS_FEATURES,
TEST_NUM_CDS_FEATURES), ""
]
assert lines[0:len(expected_metadata)] == expected_metadata,\
"Unexpected Metadata header"
lines = lines[len(expected_metadata):]
expected_summary = ["Issue summary:", "{}\t{}".format("Issue", "Count")]
assert lines[0:len(expected_summary)] == expected_summary,\
"Unexpected Issue summary header"
lines = lines[len(expected_summary):]
counts = check_fasta_gff.count_issues(TEST_CHECK_ISSUES)
expected_counts = [
"{}\t{}".format(issue, count) for issue, count in counts
]
actual_counts = lines[0:len(expected_counts)]
# Sort so test passes if same counts are reported but in
# different order.
expected_counts.sort()
actual_counts.sort()
assert actual_counts == expected_counts, "Unexpected counts"
lines = lines[len(expected_counts):]
line = lines.pop(0)
assert line == "", "Missing blank line"
line = lines.pop(0)
assert line == "Issue details:", "Unexpected Issue details header"
expected_details = [
check_fasta_gff.ISSUE_FORMATS[issue_type].format(sequence=seq_id,
feature=feature_id,
data=issue_data)
for seq_id, feature_id, issue_type, issue_data in TEST_CHECK_ISSUES
]
actual_details = lines[0:len(expected_details)]
# Sort so test passes if same details are reported but in
# different order.
expected_details.sort()
actual_details.sort()
assert actual_details == expected_details, "Unexpected details"
lines = lines[len(expected_details):]
assert lines == [""], "Missing blank line"