def fasta_read(input):
"""
fasta_read(input):
@param input can be either a file or the name of a file.
returns a list of fasta_sequence objects with all the sequences in the file.
comments (lines starting with ';') are ignored.
"""
if type(input) == str:
if input.endswith('.gz'):
import gzip
input=gzip.gzipfile(input)
else:
input=file(input)
results = []
header = ''
seq_items = []
first = true
for line in input:
if line[0] == ';':
continue # comment
elif line[0] == '>':
if not first:
seq= "".join(seq_items)
results.append(fasta_sequence(header,seq))
seq_items = []
header = line[1:-1] # eat '>' and '\n'
first = false
else:
seq_items.append(line[:-1])
if len(seq_items) > 0:
seq = "".join(seq_items)
results.append(fasta_sequence(header,seq))
return results
def fasta_write(output,s):
"""
fasta_write(output, sequence[s])
@param output either a file (opened for writing) or a filename
@param sequence it can be either a fasta_sequence or a list of fasta_sequence objects
writes the sequence(s) into the file in fasta format
"""
line_width=70
if type(output) == str:
if output.endswith('.gz'):
import gzip
output=gzip.gzipfile(output,'w')
else:
output=file(output,'w')
if type(s) == list:
for ss in s:
fasta_write(output,ss)
else:
output.write("> %s\n" % s.header)
for i in xrange(0,len(s.seq),line_width):
output.write("%s\n" % s.seq[i:i+line_width])
