def make_backbone_blast_db(project_dir, blast_db_seq, dbtype):
'It formats a blastdb when need it'
logger = logging.getLogger(LOGGER_NAME)
#the name should be the basename of the blast_db_seq
db_dir = join(project_dir, BACKBONE_DIRECTORIES['blast_databases'])
if not exists(db_dir):
makedirs(db_dir)
db_seq_fpath = join(db_dir, _get_basename(blast_db_seq))
if not exists(db_seq_fpath):
#which is the name of the new databae?
blast_db_seq_format = guess_seq_file_format(open(blast_db_seq))
if blast_db_seq_format == 'fasta':
rel_symlink(blast_db_seq, db_seq_fpath)
else:
seqio(in_seq_fhand=open(blast_db_seq),
out_seq_fhand=open(db_seq_fpath, 'w'),
out_format='fasta')
logger.info('Formatting the database %s' % db_seq_fpath)
try:
makeblastdb_plus(db_seq_fpath, dbtype=dbtype)
except RuntimeError:
msg = 'Error making blastdb. db:%s\n dbtype:%s\n' % \
(db_seq_fpath, dbtype)
remove(db_seq_fpath)
raise RuntimeError(msg)
return db_seq_fpath
def _select_last_analysis(self):
'''It select the last analysis from a directory with timestamped results'''
self._log({'analysis_started':True})
dirs = self._get_timestamped_output_dirs()
latest_dir = dirs[-1]
output_dir = self._get_output_dirs()['result']
if os.path.exists(output_dir):
os.remove(output_dir)
rel_symlink(latest_dir, output_dir)
def run(self):
'''It runs the analysis.'''
contigs_path = self._get_input_fpaths()['contigs']
contigs_ext = contigs_path.extension
reference_dir = self._create_output_dirs()['result']
reference_fpath = os.path.join(reference_dir,
BACKBONE_BASENAMES['mapping_reference'] + '.' + \
contigs_ext)
if os.path.exists(reference_fpath):
os.remove(reference_fpath)
rel_symlink(contigs_path.last_version, reference_fpath)
def run(self):
'''It runs the analysis. It checks if the analysis is already done per
input file'''
self._log({'analysis_started':True})
proj_name = self._get_project_name()
#we need an assembly dir for this run
#in this case every assembly is done in a new directory
#the directory for this assembly
output_dirs = self._create_output_dirs(timestamped=True)
assembly_dir = output_dirs['analysis']
mira_dir = os.path.join(assembly_dir, '%s_assembly' % proj_name)
original_dir = os.getcwd()
#with technologies have we used?
#each input file should have a link in the assembly dir
techs = set()
for path in self._get_input_fpaths()['reads']:
fpath = path.last_version
fname = os.path.split(fpath)[-1]
mira_input_fname = os.path.join(assembly_dir, fname)
rel_symlink(fpath, mira_input_fname)
if '454' in fname:
techs.add('454')
elif 'sanger' in fname:
techs.add('sanger')
settings = self._project_settings['Mira']
#job part of the command
job = settings['job_options']
job.extend(techs)
job = ','.join(job)
job = '-job=' + job
cmd = ['mira', '-project=%s' % proj_name, '-fasta', job]
if 'general_settings' in settings:
general_settings = settings['general_settings']
cmd.extend(general_settings)
cmd.append('-OUT:rld=yes')
for tech in techs:
tech_str = '%s_settings' % tech
if tech_str in settings:
cmd.append(tech_str.upper())
cmd.extend(settings[tech_str])
logger = logging.getLogger("franklin")
logger.info('The input files for Mira has been created')
logger.info('To run Mira go to the directory: %s' % assembly_dir)
logger.info('and run the command: %s' % ' '.join(cmd))
self._log({'analysis_finished':True})
def _prepare_mapper_index(self, mapping_index_dir, reference_path, platform,
mapper):
'It creates reference_fpath depending on the mapper and the platform'
kind = 'color' if platform == 'solid' else 'sequence'
color_space = True if kind == 'color' else False
mapping_index_dir = mapping_index_dir[0].original_path
index_dir = mapping_index_dir % (mapper, kind)
if not os.path.exists(index_dir):
os.mkdir(index_dir)
reference_fpath = reference_path.last_version
reference_fname = os.path.basename(reference_fpath)
index_fpath = os.path.join(index_dir, reference_fname)
if not os.path.exists(index_fpath):
rel_symlink(reference_fpath, index_fpath)
return index_fpath, color_space
def test_rel_symlink(self):
'It tests various cases of rel symlinks'
tempdir = NamedTemporaryDir()
hola = os.path.join(tempdir.name, 'hola')
os.mkdir (hola)
caracola = os.path.join(tempdir.name, 'caracola')
rel_symlink(hola ,caracola)
assert os.path.exists(caracola)
fname = os.path.join(hola, 'fname')
open(fname, 'w')
caracola2 = os.path.join(tempdir.name, 'caracola2')
rel_symlink(fname ,caracola2)
assert os.path.exists(caracola2)
path2 = os.path.join(tempdir.name, 'dir1', 'dir2')
os.makedirs(path2)
caracola3 = os.path.join(path2, 'caracola3')
rel_symlink(hola, caracola3)
assert os.path.exists(caracola3)
