def codon_usage_for(seq, strand = 1, frame = 0):
"""
Return a count list of codon usage for a given gene.
"""
hist = empty_codon_dict()
for codon in codons_for_frame(seq, strand, frame):
if 'N' not in codon:
hist[codon] += 1
return hist
def weighted_codon_usage_for(gene, seq, expression):
"""
Return a count list of codon usage for a given gene, weighted by the
expression base mean of the gene.
"""
if expression == 0:
return empty_codon_dict()
hist = codon_usage_for(seq)
return dict((codon, freq * expression)
for codon, freq in hist.items())
def histogram(transcripts, genes):
"""
Calculate the weighted frequencies of all occurring in the transcripts.
The transcripts are weighed by their transcription level.
Returns a dictionary of codons and their frequency.
"""
def add_to_total(total, hist):
for key, value in hist.items():
total[key] += value
total = empty_codon_dict()
failures = 0
for expression in genes:
gene = expression[0]
seq = transcripts[gene]
if seq is None or seq == '':
failures += 1
continue
hist = weighted_codon_usage_for(gene, seq, expression[1])
add_to_total(total, hist)
return total
