def ensure_pickle_goodness(pickle_filename, root_dir, fasta_files_to_add=None):
"""
Old versions of IceIterative.write_pickle is missing some key/values.
Add if needed.
Return a good pickle object, and the pickle's filename
"""
a = load(open(pickle_filename))
if realpath(a['fasta_filename']) != current_fasta(root_dir):
raise ValueError(
"The pickle file %s indicates that " % pickle_filename +
"current.fasta is not being used. ICE likely did not " +
"finish to a point that could be picked up.")
a['newids'] = check_n_fix_newids(a)
if fasta_files_to_add is not None:
for f in fasta_files_to_add.split(','):
if not op.exists(f):
raise IOError("%s is not a valid fasta file to add!" % f)
if f in a['fasta_filenames_to_add']:
log.warning("%s is already in to-add list. Ignore.", f)
a['fasta_filenames_to_add'].append(f)
if 'root_dir' not in a:
log.warning("Pickle %s missing some key-values. Fixing it.",
pickle_filename)
a['root_dir'] = root_dir
a['all_fasta_filename'] = a['all_fasta_fiilename']
a['qv_prob_threshold'] = 0.03
fixed_pickle_filename = pickle_filename + ".fixed"
with open(fixed_pickle_filename, 'w') as f:
dump(a, f)
log.info("Fixed pickle written to %s", fixed_pickle_filename)
return a, fixed_pickle_filename
else:
# newid might have been fixed, STILL output pickle writing anyway
with open(pickle_filename, 'w') as f:
dump(a, f)
return a, pickle_filename
def current_fastq(root_dir):
"""Return filename of root_dir/current.fastq, which is used in
the `current` ICE iteration."""
return realpath(op.join(root_dir, "current.fastq"))