def test_parse_cdhit_clstr_file(self):
"""parse_cdhit_clstr_file returns the correct clusters"""
data = cdhit_clstr_file.split('\n')
exp = [['seq0'],['seq1','seq10','seq3','seq23','seq145'],\
['seq7','seq17','seq69','seq1231']]
obs = parse_cdhit_clstr_file(data)
self.assertEqual(obs, exp)
def test_parse_cdhit_clstr_file(self):
"""parse_cdhit_clstr_file returns the correct clusters"""
data = cdhit_clstr_file.split('\n')
exp = [['seq0'],['seq1','seq10','seq3','seq23','seq145'],\
['seq7','seq17','seq69','seq1231']]
obs = parse_cdhit_clstr_file(data)
self.assertEqual(obs, exp)
def parse_cdhit_clusters(cfname,
output=None,
cull_empty=False,
plate_labels=True,
**kwargs):
print >> sys.stderr, 'Parsing Cluster file...'
if isinstance(cfname, file):
cfhandle = cfname
else:
cfhandle = open(os.path.join(cfname)).readlines()
clines = [line.rstrip() for line in cfhandle]
clstrs = cd_hit.parse_cdhit_clstr_file(clines)
#cull clusters with one member
if cull_empty: clstrs = filter(lambda i: len(i) > 1, clstrs)
#print 2 formats: Seq\tWell Name\tLibrary Member
# Library Member\tWell1\tWell2\tetc...
#if well member follows regex (\d+-*\d*)-(\d+), it is a well
queries = {}
lib_mems = {}
for clstr in clstrs:
#NOTE: this assumes that second DB will always come first,
#unsure if CD-HIT does this cannonically...(IT DOESNT)
lib_mem = clstr.pop(0)
lib_mems[lib_mem] = []
for member in clstr:
if plate_labels:
sanger_pos = sample.parse_sanger_id(member)
queries[sanger_pos] = lib_mem
lib_mems[lib_mem].append(sanger_pos)
else:
queries[member] = lib_mem
lib_mems[lib_mem].append(member)
return queries, lib_mems
def detect_high_homology_pairs(self, seq_list):
"""Given a list of sequences, performs an all-to-all alignment
and returns pairs of sequences (as indexes in the original list),
which have a higher than desired homology.
Args:
seq_list: List of sequence strings. Order matters in that the
pairs of indeces returned are relative to the ordering of the
list.
Returns:
A List<Set<int>> of indeces that corresponding to sequences that
were found to have homology above the given threshold with each
other.
"""
# Make sure the tmp directory
if not os.path.exists(TMP_DATA_DIR):
os.mkdir(TMP_DATA_DIR)
input_filename = os.path.join(TMP_DATA_DIR, 'test_cdhit_input.fasta')
output_prefix = os.path.join(TMP_DATA_DIR, 'test_cdhit_output')
# Write the list of sequences to a file in .fasta format, as expected
# by the cd-hit command.
with open(input_filename, 'w') as input_fh:
for seq_index in range(len(seq_list)):
# For each sequence write the following two lines.
# >0
# ATGAGATAGTA
seq = seq_list[seq_index]
input_fh.write('>' + str(seq_index) + '\n' + str(seq) + '\n')
# Call cd-hit.
cmd = [
'cd-hit',
'-i',
input_filename,
'-o',
output_prefix,
'-c',
str(self.threshold),
'-n',
str(self.word_size),
]
if self.debug:
stdout = None
else:
devnull = open(os.devnull, 'wb')
stdout = devnull
subprocess.call(cmd, stdout=stdout)
# Parse the results.
cluster_output_filename = output_prefix + '.clstr'
with open(cluster_output_filename) as cluster_output_fh:
cluster_lines = [line.rstrip() for line in cluster_output_fh]
cluster_list = cogent_cd_hit_parser.parse_cdhit_clstr_file(
cluster_lines)
# The homology conflicts are clusters with more than one element.
homology_clusters = []
for cluster in cluster_list:
if len(cluster) > 1:
homology_clusters.append(
set([int(seq_index_str) for seq_index_str in cluster]))
return homology_clusters
