def do(assemblies, labels, downloaded_dirpath, corrected_dirpath, ref_txt_fpath=None):
logger.print_timestamp()
err_fpath = os.path.join(downloaded_dirpath, 'blast.err')
blast_check_fpath = os.path.join(downloaded_dirpath, 'blast.check')
blast_res_fpath = os.path.join(downloaded_dirpath, 'blast.res')
files_md5 = dict((assembly.fpath, md5(assembly.fpath)) for assembly in assemblies)
assemblies_fpaths = dict((assembly.fpath, assembly) for assembly in assemblies)
blast_assemblies, downloaded_organisms, not_founded_organisms = \
check_blast(blast_check_fpath, blast_res_fpath, files_md5, assemblies_fpaths, assemblies, labels)
species_list = []
replacement_list = None
if ref_txt_fpath:
species_list = parse_refs_list(ref_txt_fpath)
species_by_assembly = None
else:
species_scores, species_by_assembly, replacement_dict = process_blast(blast_assemblies, downloaded_dirpath,
corrected_dirpath, labels, blast_check_fpath, err_fpath)
if species_scores:
species_scores = sorted(species_scores, reverse=True)
species_list = [species for (species, query_id, score) in species_scores]
replacement_list = [replacement_dict[query_id] for (species, query_id, score) in species_scores]
downloaded_ref_fpaths = [os.path.join(downloaded_dirpath, file) for (path, dirs, files) in os.walk(downloaded_dirpath)
for file in files if qutils.check_is_fasta_file(file)]
ref_fpaths = search_references(species_list, assemblies, labels, downloaded_dirpath, not_founded_organisms, downloaded_ref_fpaths,
blast_check_fpath, err_fpath, species_by_assembly, replacement_list)
if not ref_fpaths:
logger.main_info('Reference genomes are not found.')
if not qconfig.debug and os.path.exists(err_fpath):
os.remove(err_fpath)
ref_fpaths.sort()
return ref_fpaths
def do(assemblies, labels, downloaded_dirpath, corrected_dirpath, ref_txt_fpath=None):
logger.print_timestamp()
err_fpath = os.path.join(downloaded_dirpath, 'blast.err')
blast_check_fpath = os.path.join(downloaded_dirpath, 'blast.check')
blast_res_fpath = os.path.join(downloaded_dirpath, 'blast.res')
files_sizes = dict((assembly.fpath, os.path.getsize(assembly.fpath)) for assembly in assemblies)
assemblies_fpaths = dict((assembly.fpath, assembly) for assembly in assemblies)
blast_assemblies, downloaded_organisms, not_founded_organisms = \
check_blast(blast_check_fpath, blast_res_fpath, files_sizes, assemblies_fpaths, assemblies, labels)
organisms = []
if ref_txt_fpath:
organisms = parse_refs_list(ref_txt_fpath)
organisms_assemblies = None
else:
scores_organisms, organisms_assemblies = process_blast(blast_assemblies, downloaded_dirpath, corrected_dirpath,
labels, blast_check_fpath, err_fpath)
if scores_organisms:
scores_organisms = sorted(scores_organisms, reverse=True)
organisms = [organism for (score, organism) in scores_organisms]
downloaded_ref_fpaths = [os.path.join(downloaded_dirpath, file) for (path, dirs, files) in os.walk(downloaded_dirpath)
for file in files if qutils.check_is_fasta_file(file)]
ref_fpaths = process_refs(organisms, assemblies, labels, downloaded_dirpath, not_founded_organisms, downloaded_ref_fpaths,
blast_check_fpath, err_fpath, organisms_assemblies)
if not ref_fpaths:
logger.main_info('Reference genomes are not found.')
if not qconfig.debug and os.path.exists(err_fpath):
os.remove(err_fpath)
ref_fpaths.sort()
return ref_fpaths
def do(assemblies, labels, downloaded_dirpath, corrected_dirpath, ref_txt_fpath=None):
logger.print_timestamp()
err_fpath = os.path.join(downloaded_dirpath, 'blast.err')
blast_check_fpath = os.path.join(downloaded_dirpath, 'blast.check')
blast_res_fpath = os.path.join(downloaded_dirpath, 'blast.res')
files_md5 = dict((assembly.fpath, md5(assembly.fpath)) for assembly in assemblies)
assemblies_fpaths = dict((assembly.fpath, assembly) for assembly in assemblies)
blast_assemblies, downloaded_organisms, not_founded_organisms = \
check_blast(blast_check_fpath, blast_res_fpath, files_md5, assemblies_fpaths, assemblies, labels)
organisms = []
if ref_txt_fpath:
organisms = parse_refs_list(ref_txt_fpath)
organisms_assemblies = None
else:
scores_organisms, organisms_assemblies = process_blast(blast_assemblies, downloaded_dirpath, corrected_dirpath,
labels, blast_check_fpath, err_fpath)
if scores_organisms:
scores_organisms = sorted(scores_organisms, reverse=True)
organisms = [organism for (score, organism) in scores_organisms]
downloaded_ref_fpaths = [os.path.join(downloaded_dirpath, file) for (path, dirs, files) in os.walk(downloaded_dirpath)
for file in files if qutils.check_is_fasta_file(file)]
ref_fpaths = process_refs(organisms, assemblies, labels, downloaded_dirpath, not_founded_organisms, downloaded_ref_fpaths,
blast_check_fpath, err_fpath, organisms_assemblies)
if not ref_fpaths:
logger.main_info('Reference genomes are not found.')
if not qconfig.debug and os.path.exists(err_fpath):
os.remove(err_fpath)
ref_fpaths.sort()
return ref_fpaths
def do(assemblies, labels, downloaded_dirpath, corrected_dirpath, ref_txt_fpath=None):
logger.print_timestamp()
err_fpath = os.path.join(downloaded_dirpath, 'blast.err')
blast_check_fpath = os.path.join(downloaded_dirpath, 'blast.check')
blast_res_fpath = os.path.join(downloaded_dirpath, 'blast.res')
files_sizes = dict((assembly.fpath, os.path.getsize(assembly.fpath)) for assembly in assemblies)
assemblies_fpaths = dict((assembly.fpath, assembly) for assembly in assemblies)
blast_assemblies, downloaded_organisms, not_founded_organisms = \
check_blast(blast_check_fpath, blast_res_fpath, files_sizes, assemblies_fpaths, assemblies, labels)
species_list = []
replacement_list = None
if ref_txt_fpath:
species_list = parse_refs_list(ref_txt_fpath)
species_by_assembly = None
else:
species_scores, species_by_assembly, replacement_dict = process_blast(blast_assemblies, downloaded_dirpath,
corrected_dirpath, labels, blast_check_fpath, err_fpath)
if species_scores:
species_scores = sorted(species_scores, reverse=True)
species_list = [species for (species, query_id, score) in species_scores]
replacement_list = [replacement_dict[query_id] for (species, query_id, score) in species_scores]
downloaded_ref_fpaths = [os.path.join(downloaded_dirpath, file) for (path, dirs, files) in os.walk(downloaded_dirpath)
for file in files if qutils.check_is_fasta_file(file)]
ref_fpaths = search_references(species_list, assemblies, labels, downloaded_dirpath, not_founded_organisms, downloaded_ref_fpaths,
blast_check_fpath, err_fpath, species_by_assembly, replacement_list)
if not ref_fpaths:
logger.main_info('Reference genomes are not found.')
if not qconfig.debug and os.path.exists(err_fpath):
os.remove(err_fpath)
ref_fpaths.sort()
return ref_fpaths
def parse_meta_references(option, opt_str, value, parser, logger):
ref_fpaths = []
ref_values = value.split(',')
for i, ref_value in enumerate(ref_values):
if os.path.isdir(ref_value):
references = [join(path, file) for (path, dirs, files) in os.walk(ref_value) for file in files
if qutils.check_is_fasta_file(file, logger=logger)]
ref_fpaths.extend(sorted(references))
else:
assert_file_exists(ref_value, 'reference')
ref_fpaths.append(ref_value)
ensure_value(qconfig, option.dest, []).extend(ref_fpaths)
def parse_meta_references(option, opt_str, value, parser, logger):
ref_fpaths = []
ref_values = value.split(',')
for i, ref_value in enumerate(ref_values):
if os.path.isdir(ref_value):
references = [join(path, file) for (path, dirs, files) in os.walk(ref_value) for file in files
if qutils.check_is_fasta_file(file, logger=logger)]
ref_fpaths.extend(sorted(references))
else:
assert_file_exists(ref_value, 'reference')
ref_fpaths.append(ref_value)
ensure_value(qconfig, option.dest, []).extend(ref_fpaths)