def test_filter_table_samples(self):
"""_filter_table_samples removes small samples from OTU table
"""
otu_table = """# QIIME v%s OTU table\n#OTU ID\tsample1\tsample2\tsample3
0\t0\t2\t0
1\t1\t0\t0
2\t1\t1\t1""" % __version__
otu_table = otu_table.split('\n')
result = _filter_table_samples(otu_table, 2)
self.assertEqual(result, "# QIIME v%s OTU table\n#OTU ID\tsample1\tsample2\n0\t0\t2\n1\t1\t0\n2\t1\t1" % __version__)
result = _filter_table_samples(otu_table, 1)
self.assertEqual(result, '\n'.join(otu_table))
result = _filter_table_samples(otu_table, 3)
self.assertEqual(result, "# QIIME v%s OTU table\n#OTU ID\tsample2\n0\t2\n1\t0\n2\t1" % __version__)
def test_filter_table_samples(self):
"""_filter_table_samples removes small samples from OTU table
"""
otu_table = """#Full OTU Counts
#OTU ID\tsample1\tsample2\tsample3
0\t0\t2\t0
1\t1\t0\t0
2\t1\t1\t1""".split('\n')
result = _filter_table_samples(otu_table, 2)
self.assertEqual(result, "#Full OTU Counts\n#OTU ID\tsample1\tsample2\n0\t0\t2\n1\t1\t0\n2\t1\t1")
result = _filter_table_samples(otu_table, 1)
self.assertEqual(result, '\n'.join(otu_table))
result = _filter_table_samples(otu_table, 3)
self.assertEqual(result, "#Full OTU Counts\n#OTU ID\tsample2\n0\t2\n1\t0\n2\t1")
def test_filter_table_samples(self):
"""_filter_table_samples removes small samples from OTU table
"""
otu_table = """#Full OTU Counts
#OTU ID\tsample1\tsample2\tsample3
0\t0\t2\t0
1\t1\t0\t0
2\t1\t1\t1""".split('\n')
result = _filter_table_samples(otu_table, 2)
self.assertEqual(
result,
"#Full OTU Counts\n#OTU ID\tsample1\tsample2\n0\t0\t2\n1\t1\t0\n2\t1\t1"
)
result = _filter_table_samples(otu_table, 1)
self.assertEqual(result, '\n'.join(otu_table))
result = _filter_table_samples(otu_table, 3)
self.assertEqual(
result, "#Full OTU Counts\n#OTU ID\tsample2\n0\t2\n1\t0\n2\t1")
def main():
option_parser, opts, args = parse_command_line_parameters(**script_info)
# process options (was originally process_options())
filepath=opts.otu_table_fp
filename=filepath.strip().split('/')[-1]
filename=filename.split('.')[0]
params={}
params['otu_file'] = opts.otu_table_fp
params['min_otu_count'] = opts.min_count
params['min_otu_samples'] = opts.min_samples
params['seqs_per_sample']=opts.seqs_per_sample
seqs_per_sample=params['seqs_per_sample']
otu_file=open(params['otu_file'], 'U')
filtered_table_path=open(opts.output_otu_table_fp, 'w')
if opts.include_taxonomy:
included_taxa = set(map(strip, split_tax(opts.include_taxonomy)))
else:
included_taxa = set()
if opts.exclude_taxonomy:
excluded_taxa = set(map(strip, split_tax(opts.exclude_taxonomy)))
else:
excluded_taxa=set()
params['included_taxa']=included_taxa
params['excluded_taxa']=excluded_taxa
seq_sample_output=[]
if seqs_per_sample and params['min_otu_count']==1 and \
params['min_otu_samples']==2 and opts.include_taxonomy=='' and \
opts.exclude_taxonomy=='':
otu_file2=otu_file.readlines()
filtered_table = _filter_table_samples(otu_file2,seqs_per_sample)
filtered_table_path.write(filtered_table)
filtered_table_path.close()
elif seqs_per_sample and (params['min_otu_count']<>1 or \
params['min_otu_samples']<>2 or opts.include_taxonomy<>'' or \
opts.exclude_taxonomy<>''):
raise ValueError, 'You cannot supply seqs per sample with other filtering options. These features are mutually exclusive.'
else:
filter_table(params,filtered_table_path,otu_file)
