def records_in(handle):
"""Generator function to iterate over fasta records as Sequence
objects.
"""
with handle:
header = None
for (is_header, lines) in groupby(handle, \
lambda line : line[0] == '>'):
if is_header:
header = lines.next().rstrip()[1:]
else:
if header:
seq = "".join([line.rstrip() \
for line in lines]).upper()
yield Sequence(name=header, seq=seq)
else: # skip comment section
continue
def __init__(self, fn=None):
self.seqs = []
if fn is None:
self.seqs = []
elif fn != None and os.path.exists(fn):
fi = open(fn, 'r')
rec = []
b = False
for ln in fi:
fl = ln.rstrip()
if fl.startswith(">") and b:
#sq = self.createSeq(rec)
sn, ss = "", ""
for i in rec:
if i.startswith(">"):
sn = i[1:]
else:
ss = ss + i.replace(" ", "")
sq = Sequence(sn, ss)
del rec[:]
self.seqs.append(sq)
rec.append(fl)
elif fl.startswith(">") and not b and not rec:
b = True
rec.append(fl)
elif not fl.startswith(">"):
rec.append(fl)
sn, ss = "", ""
for i in rec:
if i.startswith(">"):
sn = i[1:]
else:
ss = ss + i.replace(" ", "")
asq = AlignedSequence(sn, ss)
del rec[:]
self.aseqs.append(asq)
else:
print("File", fn, "does not exist!!")
sys.exit()