def col_consensus(col, prev_col=[], prev_char=[]):
# Count the amino acid types in this column
aa_counts = sequtils.aa_frequencies(col)
assert aa_counts, "Column is all gaps! That's not allowed."
# Take the most common residue(s)
best_char, best_score = max(aa_counts.iteritems(),
key=lambda kv: kv[1])
# Resolve ties
ties = [aa for aa in aa_counts if aa_counts[aa] == best_score]
if len(ties) > 1:
# Breaker #1: most common after the prev. consensus char
# Resolve a tied col by restricting to rows where the preceding
# char is the consensus type for that (preceding) col
if prev_char and prev_col:
mc_next = Counter(
[b for a, b in zip(prev_col, col)
if a == prev_char[0] and b in ties]
).most_common()
ties_next = [x[0] for x in mc_next
if x[1] == mc_next[0][1]]
if ties_next:
ties = ties_next
if len(ties) > 1:
# Breaker #2: lowest overall residue frequency
ties.sort(key=lambda aa: bg_freqs[aa])
best_char = ties[0]
else:
assert best_char == ties[0], \
'WTF %s != %s[0]' % (best_char, ties)
# Save values for tie-breaker #1
prev_col[:] = col
prev_char[:] = best_char
return best_char
def col_consensus(col):
col_freqs = sequtils.aa_frequencies(col)
entroper = entropy_func(col_freqs, bg_freqs)
try:
return max(col_freqs.keys(), key=entroper)
except ValueError:
# this is probably caused by gaps
# TODO: modify this to make it more easier to debug
return '-'
def col_consensus(col):
col_freqs = sequtils.aa_frequencies(col)
entroper = entropy_func(col_freqs, bg_freqs)
return max(col_freqs.keys(), key=entroper)
