def test_format_mapping_html_data_non_location_error(self):
""" Properly formats html string for mapping file errors/warnings """
header = ["SampleID", "BarcodeSequence", "LinkerPrimerSequence", "Description"]
mapping_data = [["Sample1", "AACCGGTT", "ACATATT", "Desc_1"], ["Sample2", "CCAATTGG", "ACATATT", "Desc_2"]]
errors = ["error1\t-1,-1"]
warnings = []
# Should list errors with location -1,-1 outside of table
actual_formatted_html_data = format_mapping_html_data(header, mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data, self.expected_formatted_html_data_nonloc_error)
def test_format_mapping_html_data_warnings(self):
""" Properly formats html string for mapping file errors/warnings """
header = ["SampleID", "BarcodeSequence", "LinkerPrimerSequence", "Description"]
mapping_data = [["Sample1", "AACCGGTT", "ACATATT", "Desc_1"], ["Sample2", "CCAATTGG", "ACATATT", "Desc_2"]]
errors = []
warnings = ["warning1\t2,2"]
# Should create a an warning popup in the right location
actual_formatted_html_data = format_mapping_html_data(header, mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data, self.expected_formatted_html_warnings)
def test_format_mapping_html_data(self):
""" Properly formats html string for mapping file errors/warnings """
header = ["SampleID", "BarcodeSequence", "LinkerPrimerSequence", "Description"]
mapping_data = [["Sample1", "AACCGGTT", "ACATATT", "Desc_1"], ["Sample2", "CCAATTGG", "ACATATT", "Desc_2"]]
errors = []
warnings = []
# no errors or warnings, shouldn't get any popup mouseover data
actual_formatted_html_data = format_mapping_html_data(header, mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data, self.expected_formatted_html_no_errors_warnings)
def test_format_mapping_html_data_errors(self):
""" Properly formats html string for mapping file errors/warnings """
header = ['SampleID', 'BarcodeSequence', 'LinkerPrimerSequence',
'Description']
mapping_data = [['Sample1','AACCGGTT','ACATATT','Desc_1'],
['Sample2','CCAATTGG','ACATATT','Desc_2']
]
errors = ['problem1\t1,2']
warnings = []
# Should create a an error popup in the right location
actual_formatted_html_data = format_mapping_html_data(header,
mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data,
self.expected_formatted_html_errors)
def test_format_mapping_html_data(self):
""" Properly formats html string for mapping file errors/warnings """
header = ['SampleID', 'BarcodeSequence', 'LinkerPrimerSequence',
'Description']
mapping_data = [['Sample1','AACCGGTT','ACATATT','Desc_1'],
['Sample2','CCAATTGG','ACATATT','Desc_2']
]
errors = []
warnings = []
# no errors or warnings, shouldn't get any popup mouseover data
actual_formatted_html_data = format_mapping_html_data(header,
mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data,
self.expected_formatted_html_no_errors_warnings)
def test_format_mapping_html_data_non_location_error(self):
""" Properly formats html string for mapping file errors/warnings """
header = ['SampleID', 'BarcodeSequence', 'LinkerPrimerSequence',
'Description']
mapping_data = [['Sample1', 'AACCGGTT', 'ACATATT', 'Desc_1'],
['Sample2', 'CCAATTGG', 'ACATATT', 'Desc_2']
]
errors = ['error1\t-1,-1']
warnings = []
# Should list errors with location -1,-1 outside of table
actual_formatted_html_data = format_mapping_html_data(header,
mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data,
self.expected_formatted_html_data_nonloc_error)
def test_format_mapping_html_data_warnings(self):
""" Properly formats html string for mapping file errors/warnings """
header = ['SampleID', 'BarcodeSequence', 'LinkerPrimerSequence',
'Description']
mapping_data = [['Sample1', 'AACCGGTT', 'ACATATT', 'Desc_1'],
['Sample2', 'CCAATTGG', 'ACATATT', 'Desc_2']
]
errors = []
warnings = ['warning1\t2,2']
# Should create a an warning popup in the right location
actual_formatted_html_data = format_mapping_html_data(header,
mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data,
self.expected_formatted_html_warnings)
def test_format_mapping_html_data(self):
""" Properly formats html string for mapping file errors/warnings """
header = ['SampleID', 'BarcodeSequence', 'LinkerPrimerSequence',
'Description']
mapping_data = [['Sample1', 'AACCGGTT', 'ACATATT', 'Desc_1'],
['Sample2', 'CCAATTGG', 'ACATATT', 'Desc_2']
]
errors = []
warnings = []
# no errors or warnings, shouldn't get any popup mouseover data
actual_formatted_html_data = format_mapping_html_data(header,
mapping_data, errors, warnings)
self.assertEqual(actual_formatted_html_data,
self.expected_formatted_html_no_errors_warnings)
def check_mapping_file(mapping_fp,
output_dir=".",
has_barcodes=True,
char_replace="_",
verbose=True,
variable_len_barcodes=False,
disable_primer_check=False,
added_demultiplex_field=None,
suppress_html=False):
""" Main program function for checking mapping file
Checks mapping file for errors, warnings, writes log file, html file,
and corrected mapping file.
mapping_fp: path to metadata mapping file
output_dir: output directory for log, html, corrected mapping file.
has_barcodes: If True, will test for perform barcodes test (presence,
uniqueness, valid IUPAC DNA chars).
char_replace: Character used to replace invalid characters in data
fields. SampleIDs always use periods to be MIENS compliant.
verbose: If True, a message about warnings and/or errors will be printed
to stdout.
variable_len_barcodes: If True, suppresses warnings about barcodes of
varying length.
disable_primer_check: If True, disables tests for valid primer sequences.
added_demultiplex_field: If specified, references a field in the mapping
file to use for demultiplexing. These are to be read from fasta labels
during the actual demultiplexing step. All combinations of barcodes,
primers, and the added_demultiplex_field must be unique."""
header, mapping_data, run_description, errors, warnings =\
process_id_map(open(mapping_fp, 'U'), disable_primer_check,
has_barcodes, char_replace, variable_len_barcodes,
added_demultiplex_field, strip_quotes=False, suppress_stripping=True)
if not suppress_html:
formatted_html = format_mapping_html_data(header, mapping_data,
errors, warnings)
output_html = join(output_dir +
basename(mapping_fp).replace('.txt', '') + ".html")
html_f = open(output_html, "w")
html_f.write(formatted_html)
# get QIIME directory
qiime_dir = get_qiime_project_dir()
# Write javascript file necessary for mouseover tooltips.
# move javascript file to javascript output directory
copyfile(join(qiime_dir, 'qiime', 'support_files',
'js/overlib.js'), join(output_dir, 'overlib.js'))
corrected_mapping_data = correct_mapping_data(mapping_data,
header, char_replace)
output_corrected_fp = join(output_dir +
basename(mapping_fp).replace('.txt', '') + "_corrected.txt")
write_corrected_mapping(output_corrected_fp, header, run_description,
corrected_mapping_data)
output_log_fp = join(output_dir +
basename(mapping_fp).replace('.txt', '') + ".log")
write_log_file(output_log_fp, errors, warnings)
if verbose:
if errors or warnings:
print "Errors and/or warnings detected in mapping file. Please " +\
"check the log and html file for details."
else:
print "No errors or warnings were found in mapping file."
def check_mapping_file(mapping_fp,
output_dir=".",
has_barcodes=True,
char_replace="_",
verbose=True,
variable_len_barcodes=False,
disable_primer_check=False,
added_demultiplex_field=None):
""" Main program function for checking mapping file
Checks mapping file for errors, warnings, writes log file, html file,
and corrected mapping file.
mapping_fp: path to metadata mapping file
output_dir: output directory for log, html, corrected mapping file.
has_barcodes: If True, will test for perform barcodes test (presence,
uniqueness, valid IUPAC DNA chars).
char_replace: Character used to replace invalid characters in data
fields. SampleIDs always use periods to be MIENS compliant.
verbose: If True, a message about warnings and/or errors will be printed
to stdout.
variable_len_barcodes: If True, suppresses warnings about barcodes of
varying length.
disable_primer_check: If True, disables tests for valid primer sequences.
added_demultiplex_field: If specified, references a field in the mapping
file to use for demultiplexing. These are to be read from fasta labels
during the actual demultiplexing step. All combinations of barcodes,
primers, and the added_demultiplex_field must be unique."""
header, mapping_data, run_description, errors, warnings =\
process_id_map(open(mapping_fp, 'U'), disable_primer_check,
has_barcodes, char_replace, variable_len_barcodes,
added_demultiplex_field)
formatted_html = format_mapping_html_data(header, mapping_data,
errors, warnings)
output_html = join(output_dir +\
basename(mapping_fp).replace('.txt', '') + ".html")
html_f = open(output_html, "w")
html_f.write(formatted_html)
#get QIIME directory
qiime_dir=get_qiime_project_dir()
# Write javascript file necessary for mouseover tooltips.
# move javascript file to javascript output directory
copyfile(join(qiime_dir,'qiime','support_files',\
'js/overlib.js'), join(output_dir,'overlib.js'))
corrected_mapping_data = correct_mapping_data(mapping_data,
header, char_replace)
output_corrected_fp = join(output_dir +\
basename(mapping_fp).replace('.txt', '') + "_corrected.txt")
write_corrected_mapping(output_corrected_fp, header, run_description,
corrected_mapping_data)
output_log_fp = join(output_dir +\
basename(mapping_fp).replace('.txt', '') + ".log")
write_log_file(output_log_fp, errors, warnings)
if verbose:
if errors or warnings:
print "Errors and/or warnings detected in mapping file. Please "+\
"check the log and html file for details."
else:
print "No errors or warnings were found in mapping file."
