CodonAdaptationIndex division by zero error

[cswarth]

CodonUsage.CodonAdaptationIndex is broken in a couple ways. First, the generate_index() method will crash with a float division by zero error unless the fasta file uses every possible codon at least once.

from Bio.SeqUtils import CodonUsage
from Bio import SeqIO
from Bio.Seq import Seq
from Bio.SeqRecord import SeqRecord

with open("minimal.fa", "w") as fh:
    seq = Seq("AGTACACTGCGT")
    rec = SeqRecord(seq, id="AC12345")
    SeqIO.write(rec, fh, "fasta")

ci = CodonUsage.CodonAdaptationIndex()
ci.generate_index("minimal.fa")

produces,

$  python bin/bug.py 
Traceback (most recent call last):
  File "bin/bug.py", line 13, in <module>
    ci.generate_index("minimal.fa")
  File "/home/cwarth/.conda/envs/biopython/lib/python2.7/site-packages/Bio/SeqUtils/CodonUsage.py", line 105, in generate_index
    rcsu.append(self.codon_count[codon] / denominator)
ZeroDivisionError: float division by zero

Second, the signature of generate_index() should take a file handle to read, NOT the string name of a fasta file. Mocking a file handle for testing is a lot easier than writing a mock file to the filesystem.

[peterjc]

Taking a handle OR a filename (to preserve backward compatibility, but also matching the typical usecase) would be a sensible improvement. If you would like to submit this as a pull request, please use:

from Bio.File import as_handle

for consistency.

[peterjc]

See also #243.

[peterjc]

This is a duplicate of #243.