def move_out_chimeric(self):
chimeric_ids = self.get_chimeric_ids()
for idx_key in self.input_file_names:
fasta_file_path = os.path.join(self.indir, self.input_file_names[idx_key])
read_fasta = fa.ReadFasta(fasta_file_path)
read_fasta.close()
non_chimeric_file = fasta_file_path + self.nonchimeric_suffix
non_chimeric_fasta = fa.FastaOutput(non_chimeric_file)
fasta = fa.SequenceSource(fasta_file_path, lazy_init = False)
while fasta.next():
if not fasta.id in chimeric_ids:
non_chimeric_fasta.store(fasta, store_frequencies = False)
non_chimeric_fasta.close()
def get_chimeric_ids(self):
ids = set()
chimera_file_names = self.get_chimera_file_names(self.outdir)
file_ratio = self.check_chimeric_stats()
for file_name in chimera_file_names:
# print "from get_chimeric_ids: file_name = %s" % file_name
if file_name.endswith(self.chimeric_suffix):
both_or_denovo = self.get_chimeras_suffix(file_ratio, file_name)
# TODO: run ones for each file_base = ".".join(file_name.split(".")[0:3]) (for txt and db)
if file_name.endswith(both_or_denovo):
file_name_path = os.path.join(self.outdir, file_name)
self.utils.print_both("Get ids from %s" % file_name_path)
read_fasta = fa.ReadFasta(file_name_path)
ids.update(set(read_fasta.ids))
return ids
start_dir = args.start_dir
# start_dir = sys.argv[1]
if (is_verbatim):
print "Start from %s" % start_dir
print "Getting file names"
fa_files = seq_len.get_files(start_dir, args.ext)
if (is_verbatim):
print "Found %s fa files" % (len(fa_files))
for file_name in fa_files:
if (is_verbatim):
print file_name
try:
f_input = fa.ReadFasta(file_name)
if (args.short_s):
seq_len.print_short_seq(f_input, file_name, args.min_len)
if (args.histogram):
seq_len.get_seq_len_distrib(f_input)
except RuntimeError:
if (is_verbatim):
print sys.exc_info()[0]
except:
print "Unexpected error:", sys.exc_info()[0]
raise
next
if (is_verbatim):
print "Current directory:"