def get_abundance_from_fasta(fasta, regexp=r'count=(\d+\.\d+|\d+)'):
abundance = {}
in_fasta_handler = open(fasta, 'r')
p = re.compile(regexp)
for i, [header, _] in enumerate(read_fasta_file_handle(in_fasta_handler)):
id = header.split()[0].strip()
m = re.search(p, header)
if not m:
logger.fatal("Can't retrieve abundance information:\nfasta_path:%s\nfasta_header:%s" % (fasta, header))
sys.exit("Cant't retrieve abundance")
abundance[id] = float(m.group(1))
in_fasta_handler.close()
return abundance
def complete_fasta_with_abundance(input_fasta, output_fasta, abundance):
in_fasta_handler = open(input_fasta, 'r')
out_fasta_handler = open(output_fasta, 'w')
for header, seq in read_fasta_file_handle(in_fasta_handler):
id = header.split()[0].strip()
ab = 0
if id not in abundance:
logger.warning("Can't find the abundance for:%s. Set to 0 by default." % id)
else:
ab = abundance[id]
header = '{header} count={abundance}'.format(header=header, abundance=ab)
out_fasta_handler.write( '>{header}\n{seq}\n'.format(header=header, seq=format_seq(seq)) )
in_fasta_handler.close()
out_fasta_handler.close()