def make_gaps_consistent(self):
mainseq = self.master.seq
for i in xrange(1,len(self)):
entry = self[i]
seq = array('c', entry.seq)
for i,c in enumerate(mainseq):
if isGap(c):
seq[i] = '-'
elif isGap(seq[i]):
seq[i] = '?'
entry.seq = seq.tostring()
def remove_gaps(self, template_seq=None):
"""remove_gaps([template_seq]) : Removes all alignment columns containing gaps.
Only applicable if sequences are aligned (at least within each EntryGroup).
If template_seq is given, all entries must be aligned to template_seq.
"""
# Note that using this function will unalign entries with different codes!
if template_seq:
assert isinstance(template_seq[0], str)
for eg in self:
code = eg.code
if template_seq:
template = template_seq
else:
if not eg.hasMasterEntry():
continue
template = eg[0].seq
seqs = []
for entry in eg:
assert len(entry.seq) == len(template)
seqs.append(array('c', entry.seq))
iend = len(template)
ingap = False
for i in xrange(len(template), 0, -1):
c = template[i-1]
#print i,c,ingap
if ingap:
if not isGap(c):
ingap = False
for seq in seqs:
del seq[i:iend]
else:
if isGap(c):
ingap = True
iend = i
if ingap:
#print "removing final gap"
for seq in seqs:
del seq[:iend]
for i,entry in enumerate(eg):
entry.seq = seqs[i].tostring()
def remove_gapped_columns(seqs, master=0):
# Keep only those columns where master sequence has no gap
#
residuesToDelete = []
n = 0
for i,c in enumerate(seqs[master]):
if isGap(c):
residuesToDelete.append(i)
for i in xrange(len(seqs)):
assert len(seqs[i]) == len(seqs[-1]), "Sequences not aligned"
seqs[i] = __stringRemoveIndeces(seqs[i], residuesToDelete)
def find_aligned_residues(seq1, seq2, equiv=None, eqchars=":."):
assert len(seq1) == len(seq2)
assert None == equiv or len(seq1) == len(equiv)
numbering_list1 = []
numbering_list2 = []
n1=0
n2=0
if None == equiv:
for i in xrange(len(seq1)):
if not isGap(seq1[i]) and not isGap(seq2[i]):
numbering_list1.append(n1)
numbering_list2.append(n2)
n1 += 1
n2 += 1
elif not isGap(seq1[i]):
n1 += 1
elif not isGap(seq2[i]):
n2 += 1
else:
for i, e in enumerate(equiv):
if equiv[i] in eqchars:
numbering_list1.append(n1)
numbering_list2.append(n2)
if not isGap(seq1[i]):
n1 += 1
if not isGap(seq2[i]):
n2 += 1
return numbering_list1, numbering_list2
def __get_pid_counts(seq1, seq2):
assert len(seq1) == len(seq2), "\n>seq1\n%s\n>seq2\n%s\n"%(seq1, seq2)
l1 = 0 # ungapped length of seq1
l2 = 0 # ungapped length of seq2
cov = 0 # number of aligned pairs (absolute coverage)
idn = 0 # number of identical pairs (absolute identity)
for i in xrange(len(seq1)):
c1 = seq1[i]
c2 = seq2[i]
if (not isGap(c1)):
l1 += 1
if (not isGap(c2)):
l2 += 1
cov += 1
if c1 == c2:
idn += 1
elif (not isGap(c2)):
l2 += 1
return (l1, l2, cov, idn)
def consensus(seqs):
cons=[]
totals=[]
length = len(seqs[0])
for s in seqs:
assert len(s) == length, "Input sequences must be aligned (of the same length)"
for i in xrange(length):
d = {}
n=0
for j in xrange(len(seqs)):
c = seqs[j][i]
if isGap(c):
continue
n+=1
if c not in d:
d[c]=1
else:
d[c]+=1
residues=[]
for res in sorted(d):
residues.append((res, d[res]))
cons.append(residues)
totals.append(n)
return cons, totals