def collect_proteomes_and_annotaitons(input_dir):
proteomes = []
annotations = []
files = listdir(input_dir)
if not files:
interrupt('Directory contains no files.')
for f in (join(input_dir, f) for f in files if isfile(join(input_dir, f))):
if '.' in f and splitext(f)[1] in ['.fasta', '.faa', '.fa', '.fsa']:
try:
log.debug(' Checking if %s is fasta.' % f)
next(SeqIO.parse(f, 'fasta'))
except ValueError, e:
pass
else:
proteomes.append(f)
continue
if '.' in f and splitext(f)[1] in ['.gb', '.genbank', '.gbk']:
try:
log.debug(' Checking if %s is genbank.' % f)
SeqIO.read(f, 'genbank')
except Exception, e:
log.debug(str(e) + ', ' + f)
else:
annotations.append(f)
def collect_proteomes_and_annotaitons(input_dir):
proteomes = []
annotations = []
files = listdir(input_dir)
if not files:
interrupt('Directory contains no files.')
for f in (join(input_dir, f) for f in files if isfile(join(input_dir, f))):
if '.' in f and splitext(f)[1] in ['.fasta', '.faa', '.fa', '.fsa']:
try:
log.debug(' Checking if %s is fasta.' % f)
next(SeqIO.parse(f, 'fasta'))
except ValueError, e:
pass
else:
proteomes.append(f)
continue
if '.' in f and splitext(f)[1] in ['.gb', '.genbank', '.gbk']:
try:
log.debug(' Checking if %s is genbank.' % f)
SeqIO.read(f, 'genbank')
except Exception, e:
log.debug(str(e) + ', ' + f)
else:
annotations.append(f)
def run(starting_from_here=False):
if p.species_list:
if not test_entrez_conn():
log.error(' No internet connection: cannot fetch annotations.')
return 4
log.debug(' Using species list: ' + str(p.species_list))
gb_ids = read_list(p.species_list)
log.debug('species_list: ' + str(gb_ids))
res = fetch_annotations_species_name_entrez(config.annotations_dir, gb_ids, p.proxy)
if res != 0: return res
return make_proteomes(config.annotations_dir, config.proteomes_dir)
elif p.ids_list:
if not test_entrez_conn():
log.error('No internet connection: cannot fetch annotations.')
return 4
log.debug(' Using ref ids: ' + str(p.ids_list))
ref_ids = read_list(p.ids_list)
res = fetch_annotations_for_ids(config.annotations_dir, ref_ids)
if res != 0: return res
return make_proteomes(config.annotations_dir, config.proteomes_dir)
else:
proteomes, annotations = [], []
if p.proteomes:
proteomes, annotations = collect_proteomes_and_annotaitons(p.proteomes)
if proteomes == []:
interrupt('No fasta found in ' + p.proteomes)
if p.annotations:
proteomes, annotations = collect_proteomes_and_annotaitons(p.annotations)
if annotations == []:
interrupt('No gb files found in ' + p.annotations)
#if not proteomes and not annotations:
# interrupt('Directory must contain fasta or genbank files.')
#
#if proteomes and annotations:
# log.warn('Directory %s contains both fasta and genbank files, using fasta.')
if annotations:
if not isdir(config.annotations_dir):
mkdir(config.annotations_dir)
for annotation in annotations:
copy(annotation, config.annotations_dir)
return make_proteomes(config.annotations_dir, config.proteomes_dir)
elif proteomes:
if not isdir(config.proteomes_dir):
mkdir(config.proteomes_dir)
if p.download_anno:
if not test_entrez_conn():
#log.error(' Error: no internet connection, cannot fetch annotations. '
# 'You can start without a --no-fetch option, in this case '
# 'a reduced version of orthogroups.txt with no annotations will be produced.')
#return 1
log.error(' Warning: no internet connection, cannot fetch annotations. '
'A reduced version of orthogroups.txt with no annotations will be produced.')
else:
# ref_ids = [splitext(basename(prot_file))[0] for prot_file in proteomes]
# fetch_annotations_for_ids(config.annotations_dir, ref_ids)
gb_ids = [splitext(basename(prot_file))[0] for prot_file in proteomes]
log.debug('ids_list: ' + str(gb_ids))
res = fetch_annotations_for_ids(config.annotations_dir, gb_ids, p.proxy)
if res > 0:
return res
if res == -1:
p.download_anno = False
return adjust_proteomes(proteomes, config.proteomes_dir, p.prot_id_field)
def run(starting_from_here=False):
if p.species_list:
if not test_entrez_conn():
log.error(
' No internet connection: cannot fetch annotations.')
return 4
log.debug(' Using species list: ' + str(p.species_list))
gb_ids = read_list(p.species_list)
log.debug('species_list: ' + str(gb_ids))
res = fetch_annotations_species_name_entrez(
config.annotations_dir, gb_ids, p.proxy)
if res != 0: return res
return make_proteomes(config.annotations_dir, config.proteomes_dir)
elif p.ids_list:
if not test_entrez_conn():
log.error('No internet connection: cannot fetch annotations.')
return 4
log.debug(' Using ref ids: ' + str(p.ids_list))
ref_ids = read_list(p.ids_list)
res = fetch_annotations_for_ids(config.annotations_dir, ref_ids)
if res != 0: return res
return make_proteomes(config.annotations_dir, config.proteomes_dir)
else:
proteomes, annotations = [], []
if p.proteomes:
proteomes, annotations = collect_proteomes_and_annotaitons(
p.proteomes)
if proteomes == []:
interrupt('No fasta found in ' + p.proteomes)
if p.annotations:
proteomes, annotations = collect_proteomes_and_annotaitons(
p.annotations)
if annotations == []:
interrupt('No gb files found in ' + p.annotations)
#if not proteomes and not annotations:
# interrupt('Directory must contain fasta or genbank files.')
#
#if proteomes and annotations:
# log.warn('Directory %s contains both fasta and genbank files, using fasta.')
if annotations:
if not isdir(config.annotations_dir):
mkdir(config.annotations_dir)
for annotation in annotations:
copy(annotation, config.annotations_dir)
return make_proteomes(config.annotations_dir,
config.proteomes_dir)
elif proteomes:
if not isdir(config.proteomes_dir):
mkdir(config.proteomes_dir)
if p.download_anno:
if not test_entrez_conn():
#log.error(' Error: no internet connection, cannot fetch annotations. '
# 'You can start without a --no-fetch option, in this case '
# 'a reduced version of orthogroups.txt with no annotations will be produced.')
#return 1
log.error(
' Warning: no internet connection, cannot fetch annotations. '
'A reduced version of orthogroups.txt with no annotations will be produced.'
)
else:
# ref_ids = [splitext(basename(prot_file))[0] for prot_file in proteomes]
# fetch_annotations_for_ids(config.annotations_dir, ref_ids)
gb_ids = [
splitext(basename(prot_file))[0]
for prot_file in proteomes
]
log.debug('ids_list: ' + str(gb_ids))
res = fetch_annotations_for_ids(
config.annotations_dir, gb_ids, p.proxy)
if res > 0:
return res
if res == -1:
p.download_anno = False
return adjust_proteomes(proteomes, config.proteomes_dir,
p.prot_id_field)
