def format_out(groups, amp_num, fa_amp_file_name, session_dir):
'''Format output'''
records = FastaFormatParser.parse(open(os.path.join(session_dir, fa_amp_file_name)))
seq_dict = {}
for record in records:
id = record['id'].split()[0]
desc = record['desc']
seq = record['seq']
seq = chilli.print_seq(seq, 80)
fa_seq = '>%s %s%s%s%s' % (id, desc, os.linesep, seq, os.linesep)
seq_dict[id] = fa_seq
group_list = []
for group_sn, group in enumerate(groups):
seq_list = [seq_dict[amp_id] for amp_id, strand in group]
file_name = write_amplicons(session_dir, seq_list)
file_name = os.path.join(session_dir, file_name)
if len(group) < 2:
t_coffee_result = file_name
else:
t_coffee_result = run_t_coffee(file_name)
t_coffee_result = os.path.basename(t_coffee_result)
group_list.append((group_sn+1, ', '.join(['%s (%s)' % (amp_id, strand) for amp_id, strand in group]), t_coffee_result))
return group_list
def check_infile(options):
'''Check and return Oligos'''
err_or_degenerate = SeqCheck.fasta_format_check(options.infile)
if err_or_degenerate in ['yes', 'no']:
global degenerate
degenerate = err_or_degenerate
else:
print2stderr(err_or_degenerate)
options.infile.seek(0)
if degenerate == 'no':
oligos = FastaFormatParser.parse(options.infile)
else:
oligos = DegenerateSeqConvetor.convert(FastaFormatParser.parse(options.infile))
options.infile.close()
return oligos
def check_infile(options):
'''Check and return Oligos'''
err_or_degenerate = SeqCheck.fasta_format_check(options.infile)
if err_or_degenerate in ['yes', 'no']:
global degenerate
degenerate = err_or_degenerate
else:
print2stderr(err_or_degenerate)
options.infile.seek(0)
if degenerate == 'no':
oligos = FastaFormatParser.parse(options.infile)
else:
oligos = DegenerateSeqConvetor.convert(
FastaFormatParser.parse(options.infile))
options.infile.close()
return oligos
def analysis_hairpin(infile, mv, dv, oligo, dntp, align_mode):
'''Analysis dimer'''
records = FastaFormatParser.parse(open(infile))
dimer_list = []
for i in range(len(records)):
record_i = records[i]
id_i = record_i['id']
seq_i = record_i['seq']
delta_g, tm, align = run_ntthal(seq_i, seq_i, mv=mv, dv=dv, d=oligo, n=dntp, align_mode=align_mode)
if tm >= 0:
dimer_list.append([id_i, seq_i, tm, delta_g, align])
dimer_list.sort(key=itemgetter(2), reverse=True)
return dimer_list
def get_mid_seq(mid_seq_id_list, options, session_dir, db):
'''Get the amp sequence using twoBitToFa from Blat suite'''
mid_seq_list = []
if len(mid_seq_id_list) == 0:
return mid_seq_list
tmp_name = '%s.mid_seq_id_list.txt.tmp' % (os.path.basename(db))
mid_seq_id_list_file = os.path.join(session_dir, tmp_name)
fh = open(mid_seq_id_list_file, 'w')
fh.write(os.linesep.join(mid_seq_id_list))
fh.close()
tmp_name = '%s.twoBitToFa_output.txt.tmp' % (os.path.basename(db))
outfile = os.path.join(session_dir, tmp_name)
twoBitDB = db + '.2bit'
cmd = '%s%stwoBitToFa -seqList=%s %s %s' % (
bin_path, os.sep, mid_seq_id_list_file, twoBitDB, outfile)
try:
out, err = subprocess.Popen(cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate()
except:
msg = 'Error: twoBitToFa running error.'
print msg
print2stderr(msg)
if err:
print err
print2stderr(err)
try:
fh = open(outfile)
except:
msg = 'Error: twoBitToFa running error: no output file produced.'
print msg
print2stderr(msg)
records = FastaFormatParser.parse(fh)
mid_seq_list = []
for record in records:
mid_seq_list.append(record['seq'].lower())
fh.close()
return mid_seq_list
def main():
'''Main'''
options = get_opt()
try:
fh = open(options.infile)
except:
chilli.print2stderr('Error: can not open file: %s' % (options.infile))
records = FFP.parse(fh)
result_fasta_array = convert(records)
out = format2fasta(result_fasta_array)
if options.outfile == sys.stdout:
print(out)
else:
fo = open(options.outfile, 'w')
fo.write(out)
fo.close()
def extract_by_twoBitToFa(id_pos_range_list, options, session_dir, db):
'''Get the amp sequence using twoBitToFa from Blat suite'''
seq_list = []
if len(id_pos_range_list) == 0:
return seq_list
tmp_name = '%s.id_pos_range_list.txt.tmp' % (os.path.basename(db))
id_pos_range_list_file = os.path.join(session_dir, tmp_name)
fh = open(id_pos_range_list_file, 'w')
fh.write(os.linesep.join(id_pos_range_list))
fh.close()
tmp_name = '%s.twoBitToFa_output.txt.tmp' % (os.path.basename(db))
outfile = os.path.join(session_dir, tmp_name)
twoBitDB = db + '.2bit'
cmd = 'MFEprimer_%s_%s_twoBitToFa -seqList=%s %s %s' % (platform.system(
), get_bit(), id_pos_range_list_file, twoBitDB, outfile)
try:
out, err = subprocess.Popen(cmd,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE).communicate()
except:
msg = 'Error: twoBitToFa running error.'
print2stderr(msg)
if err:
print2stderr(err)
try:
fh = open(outfile)
except:
msg = 'Error: twoBitToFa running error: no output file produced.'
print2stderr(msg)
records = FastaFormatParser.parse(fh)
seq_list = []
for record in records:
seq_list.append(record['seq'].lower())
fh.close()
return seq_list
def get_mid_seq(mid_seq_id_list, options, session_dir, db):
'''Get the amp sequence using twoBitToFa from Blat suite'''
mid_seq_list = []
if len(mid_seq_id_list) == 0:
return mid_seq_list
tmp_name = '%s.mid_seq_id_list.txt.tmp' % (os.path.basename(db))
mid_seq_id_list_file = os.path.join(session_dir, tmp_name)
fh = open(mid_seq_id_list_file, 'w')
fh.write(os.linesep.join(mid_seq_id_list))
fh.close()
tmp_name = '%s.twoBitToFa_output.txt.tmp' % (os.path.basename(db))
outfile = os.path.join(session_dir, tmp_name)
twoBitDB = db + '.2bit'
cmd = '%s%stwoBitToFa -seqList=%s %s %s' % (bin_path, os.sep, mid_seq_id_list_file, twoBitDB, outfile)
try:
out, err = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE).communicate()
except:
msg = 'Error: twoBitToFa running error.'
print msg
print2stderr(msg)
if err:
print err
print2stderr(err)
try:
fh = open(outfile)
except:
msg = 'Error: twoBitToFa running error: no output file produced.'
print msg
print2stderr(msg)
records = FastaFormatParser.parse(fh)
mid_seq_list = []
for record in records:
mid_seq_list.append(record['seq'].lower())
fh.close()
return mid_seq_list
