def fetch_annotations_for_ids(annotations_dir, ref_ids, proxy=None):
ref_ids = list(chain(*[__range_of_ref_ids(line) for line in ref_ids]))
if None in ref_ids:
return 1
if not isdir(annotations_dir):
mkdir(annotations_dir)
if proxy:
setup_http_proxy(*proxy.split(':'))
if ref_ids == []:
log.info(' No references have been found.')
ids = raw_input(' Put reference ids manually:').split()
if ids == []:
log.error(' No references :(')
return 1
log.info(' IDs: %s' % ', '.join(ref_ids))
for i, ref_id in enumerate(ref_ids):
log.info(' Fetching annotations for %s...' % ref_id)
try:
try:
fetch_handle = Entrez.efetch(db='nucleotide', id=ref_id,
retmode='text', rettype='gbwithparts')
except urllib2.HTTPError as e:
log.error(' Error: cannot fetch data for reference id ' + ref_id)
log.error(' Http error code: %s, reason: %s' % (str(e.code), str(e.reason)))
return -1
else:
gb_fpath = join(annotations_dir, ref_id + '.gb')
with open(gb_fpath, 'w') as file:
file.write(fetch_handle.read())
rec = SeqIO.read(gb_fpath, 'genbank')
genes_number = 0
cds_number = 0
for f in rec.features:
if f.type == 'gene':
genes_number += 1
if f.type == 'CDS':
cds_number += 1
log.info(' ' + rec.description)
log.info(' %d genes, %d coding regions found.' % (genes_number, cds_number))
log.info(' saved %s' % gb_fpath)
if i + 1 < len(ref_ids):
log.info('')
except KeyboardInterrupt, e:
return 1
def fetch_annotations_for_species_from_ftp(save_dir, species_names, proxy=None, clip=None):
if not species_names:
log.error(' No species names')
return 1
if proxy:
setup_http_proxy(*proxy.split(':'))
if not isdir(save_dir):
mkdir(save_dir)
ftp = FTP('ftp.ncbi.nih.gov')
log.debug(' Logging in ' + ftp.login())
ftp.cwd('genomes/Bacteria')
for i, dirname in enumerate(ftp.nlst()):
for sp_i, sp in enumerate(species_names):
sp = sp.replace(' ', '_').strip()
if sp == '' or sp[0] == '#':
continue
if dirname.startswith(sp):
log.debug(' Scanning ' + dirname)
for remote_fpath in ftp.nlst(dirname):
if remote_fpath.endswith('.gbk'):
dest_fname = str(i) + '_' + str(sp_i) + '_' + basename(remote_fpath)
dest_fpath = join(save_dir, dest_fname)
ftp.retrbinary('RETR ' + remote_fpath, open(dest_fpath, 'wb').write)
rec = SeqIO.read(dest_fpath, 'gb')
log.info(' Definition: ' + rec.description)
# if 'plasmid' in rec.description:
# remove(dest_fpath)
# else:
log.info(' saved ' + dest_fpath)
log.info('')
return 0
