def download_and_process_viral_refseq(merged_viral_faas=None,
output_dir='.',
viral_files=2,
threads=10,
verbose=True):
"""Can only download newest version"""
# download all of the viral protein files, need to know the number of files
# TODO: Make it so that you don't need to know number of viral files in refseq viral
if merged_viral_faas is None: # download database if not provided
faa_base_name = 'viral.%s.protein.faa.gz'
viral_faa_glob = path.join(output_dir, faa_base_name % '*')
for number in range(viral_files):
number += 1
refseq_url = 'ftp://ftp.ncbi.nlm.nih.gov/refseq/release/viral/viral.%s.protein.faa.gz' % number
refseq_faa = path.join(output_dir, faa_base_name % number)
download_file(refseq_url, refseq_faa, verbose=verbose)
# then merge files from above
merged_viral_faas = path.join(output_dir,
'viral.merged.protein.faa.gz')
run_process([
'cat %s > %s' % (' '.join(glob(viral_faa_glob)), merged_viral_faas)
],
shell=True)
# make mmseqs database
refseq_viral_mmseqs_db = path.join(
output_dir, 'refseq_viral.%s.mmsdb' % get_iso_date())
make_mmseqs_db(merged_viral_faas,
refseq_viral_mmseqs_db,
create_index=True,
threads=threads,
verbose=verbose)
return refseq_viral_mmseqs_db
def download_and_process_kofam_ko_list(kofam_ko_list_gz=None,
output_dir='.',
verbose=False):
if kofam_ko_list_gz is None:
kofam_ko_list_gz = path.join(output_dir, 'kofam_ko_list.tsv.gz')
download_file('ftp://ftp.genome.jp/pub/db/kofam/ko_list.gz',
kofam_ko_list_gz,
verbose=verbose)
# TODO: fix this so that it is gunzipped to the path
kofam_ko_list = path.join(output_dir, 'kofam_ko_list.tsv')
run_process(['gunzip', kofam_ko_list_gz], verbose=verbose)
return kofam_ko_list
def download_and_process_dbcan(dbcan_hmm=None,
output_dir='.',
dbcan_release='8',
verbose=True):
if dbcan_hmm is None: # download database if not provided
dbcan_hmm = path.join(output_dir,
'dbCAN-HMMdb-V%s.txt' % dbcan_release)
download_file(
'http://bcb.unl.edu/dbCAN2/download/Databases/dbCAN-HMMdb-V%s.txt'
% dbcan_release,
dbcan_hmm,
verbose=verbose)
run_process(['hmmpress', '-f', dbcan_hmm], verbose=verbose)
return dbcan_hmm
def download_and_process_vogdb(vog_hmm_targz=None,
output_dir='.',
vogdb_release='latest',
verbose=True):
if vog_hmm_targz is None:
vog_hmm_targz = path.join(output_dir, 'vog.hmm.tar.gz')
vogdb_url = 'http://fileshare.csb.univie.ac.at/vog/%s/vog.hmm.tar.gz' % vogdb_release
download_file(vogdb_url, vog_hmm_targz, verbose=verbose)
hmm_dir = path.join(output_dir, 'vogdb_hmms')
mkdir(hmm_dir)
vogdb_targz = tarfile.open(vog_hmm_targz)
vogdb_targz.extractall(hmm_dir)
vog_hmms = path.join(output_dir, 'vog_%s_hmms.txt' % vogdb_release)
merge_files(glob(path.join(hmm_dir, 'VOG*.hmm')), vog_hmms)
run_process(['hmmpress', '-f', vog_hmms], verbose=verbose)
return vog_hmms
def download_and_process_kofam_hmms(kofam_profile_tar_gz=None,
output_dir='.',
verbose=False):
if kofam_profile_tar_gz is None:
kofam_profile_tar_gz = path.join(output_dir, 'kofam_profiles.tar.gz')
download_file('ftp://ftp.genome.jp/pub/db/kofam/profiles.tar.gz',
kofam_profile_tar_gz,
verbose=verbose)
kofam_profiles = path.join(output_dir, 'kofam_profiles')
mkdir(kofam_profiles)
run_process(['tar', '-xzf', kofam_profile_tar_gz, '-C', kofam_profiles],
verbose=verbose)
merged_kofam_profiles = path.join(output_dir, 'kofam_profiles.hmm')
merge_files(glob(path.join(kofam_profiles, 'profiles', '*.hmm')),
merged_kofam_profiles)
run_process(['hmmpress', '-f', merged_kofam_profiles], verbose=verbose)
return merged_kofam_profiles
def process_mmspro(full_alignment,
output_dir,
db_name='db',
threads=10,
verbose=True):
mmseqs_msa = path.join(output_dir, '%s.mmsmsa' % db_name)
run_process(['mmseqs', 'convertmsa', full_alignment, mmseqs_msa],
verbose=verbose)
mmseqs_profile = path.join(output_dir, '%s.mmspro' % db_name)
run_process([
'mmseqs', 'msa2profile', mmseqs_msa, mmseqs_profile, '--match-mode',
'1', '--threads',
str(threads)
],
verbose=verbose)
tmp_dir = path.join(output_dir, 'tmp')
run_process([
'mmseqs', 'createindex', mmseqs_profile, tmp_dir, '-k', '5', '-s', '7',
'--threads',
str(threads)
],
verbose=verbose)
return mmseqs_profile
def test_run_process():
run_process(['echo', 'Hello', 'World'], verbose=True)
assert True
