def _generateProbabilitiesFromFrequencies(self):
# Turn frequencies into probabilities
total_codons = float(sum(self.codon_freq.values()))
self.codon_prob = dict([(codon, self.codon_freq[codon]/total_codons) for codon in self.codon_freq.keys()])
total_nucleotides = float(sum(self.nucleotide_freq.values()))
self.codon_prob_from_nucleotide = {}
for codon in self.codon_freq.keys():
cprob = math.exp(sum([math.log(self.nucleotide_freq[nt]/total_nucleotides) for nt in codon]))
self.codon_prob_from_nucleotide[codon] = cprob
self.codon_prob_given_aa = {}
for aa in translate.AAsAndStop():
codons = translate.getCodons(aa, rna=False)
#alt_codons[aa] = codons
marginal_prob = sum([self.codon_prob[c] for c in codons])
for codon in codons:
# Compute the conditional probability of a codon, given that the amino acid is specified
if marginal_prob > 0.0:
self.codon_prob_given_aa[codon] = self.codon_prob[codon]/marginal_prob
else:
self.codon_prob_given_aa[codon] = 0.0
def __str__(self):
s = ""
if not self.nucleotide_freq is None:
s += '\nnt\tnt.freq\n'
for nt in 'ATGC':
s += '{0}\t{1:d}\n'.format(nt, int(self.nucleotide_freq[nt]))
s += '\n'
if not self.codon_freq is None:
if self.codon_prob is None:
print "# generating probs from freqs"
self._generateProbabilitiesFromFrequencies()
if self.codon_syn_scores is None:
print "# generating scores from probs"
self._generateScoresFromProbabilities()
s += 'aa\tcodon\tcodon.freq\tcodon.prob\tcodon.prob.from.nt\tcodon.cond.prob.given.aa\tsyn\n'
for aa in translate.AAsAndStop():
codons = translate.getCodons(aa, rna=False)
for codon in codons:
s += '{0}\t{1}\t{2:d}\t{3:.5f}\t{4:.5f}\t{5:.5f}\t{6:.5f}\n'.format(aa, codon, int(self.codon_freq[codon]), self.codon_prob[codon], self.codon_prob_from_nucleotide[codon], self.codon_prob_given_aa[codon], self.codon_syn_scores[codon])
s += '\n'
return s
