def goldman_q_dna_pair(seq1, seq2):
"""Returns the Goldman rate matrix"""
if len(seq1) != len(seq2):
raise ValueError, "seq1 and seq2 are not the same length!"
seq1, seq2 = ModelDnaSequence(seq1), ModelDnaSequence(seq2)
m = Counts.fromPair(seq1, seq2, DnaPairs, average=True)._data
q = m / m.sum(axis=1)[:, NewAxis]
new_diag = -(q.sum(axis=1) - diag(q))
for i, v in enumerate(new_diag):
q[i, i] = v
return q
def freqs_from_aln_array(seqs):
"""Returns per-position freqs from arbitrary size alignment.
Warning: fails if all seqs aren't the same length.
written by Rob Knight
seqs = list of lines from aligned fasta file
"""
result = None
for label, seq in MinimalFastaParser(seqs):
# Currently cogent does not support . characters for gaps, converting
# to - characters for compatability.
seq = ModelDnaSequence(seq.replace('.','-'))
if result is None:
result = zeros((len(seq.Alphabet), len(seq)),dtype=int)
indices = arange(len(seq), dtype=int)
result[seq._data,indices] += 1
return Profile(result, seq.Alphabet)
def goldman_q_dna_triple(seq1, seq2, outgroup):
"""Returns the Goldman rate matrix for seq1"""
if len(seq1) != len(seq2) != len(outgroup):
raise ValueError, "seq1,seq2 and outgroup are not the same length!"
seq1 = ModelDnaSequence(seq1)
seq2 = ModelDnaSequence(seq2)
outgroup = ModelDnaSequence(outgroup)
m = Counts.fromTriple(seq1, seq2, outgroup, DnaPairs)._data
q = m / m.sum(axis=1)[:, NewAxis]
new_diag = -(q.sum(axis=1) - diag(q))
for i, v in enumerate(new_diag):
q[i, i] = v
return q