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(start_from_here=False):
assemblies = [
join(p.assemblies, f) for f in listdir(p.assemblies)
if f and f[0] != '.'
]
if isdir(config.proteomes_dir):
assemblies = filter_dublicated_proteomes(
config.proteomes_dir, assemblies)
if assemblies == []:
log.warn(all_considered_warning % config.proteomes_dir)
exit(1)
assembly_names = [splitext(basename(asm))[0] for asm in assemblies]
filtered_assemblies = [
join(assemblies_dir, asm_name + '.fna')
for asm_name in assembly_names
]
new_proteomes = [
join(config.proteomes_dir, asm_name + '.fasta')
for asm_name in assembly_names
]
if not isdir(assemblies_dir): mkdir(assemblies_dir)
log.debug(' Created assemblies_dir ' + assemblies_dir)
total_successful_filters = 0
for assembly, filtered_asm in zip(assemblies, filtered_assemblies):
if filter_assembly(assembly,
filtered_asm,
skip=(4, 7, 10, 23, 32, 38),
skip_after=51) == 0:
total_successful_filters += 1
if total_successful_filters == 0:
log.error('No correct assemblies.')
return 1
for asm, prot, asm_name in zip(filtered_assemblies, new_proteomes,
assembly_names):
res = cmdline('prodigal',
parameters=[
'-i', asm, '-o',
join(config.intermediate_dir, asm_name),
'-a', prot
])()
if res != 0:
return res
log.info('')
res = adjust_proteomes(new_proteomes,
config.proteomes_dir,
prot_id_field=0)
if res != 0:
return res
# Recreate new_proteomes_directory
if exists(new_proteomes_dir):
rmtree(new_proteomes_dir)
if not isdir(new_proteomes_dir):
mkdir(new_proteomes_dir)
for prot in new_proteomes:
copy(prot, join(new_proteomes_dir, basename(prot)))
return 0
