def _filter_out_genomes(data):
""" Filters out genomes found in run_info.yaml
"""
print "AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA"
print data
# genome_build, sam_ref = ref_genome_info(data["info"], config, data["dirs"])
sam_ref = data["sam_ref"]
log.info("Removing genome from sample %s" % str(data["name"]))
try:
# not data ! should reach run_info.yaml somehow from here
if data["filter_out_genomes"]:
for genome in data["filter_out_genomes"].split(","):
(out_file, ext) = os.path.splitext(os.path.basename(fastq1))
out_file = out_file + "-stripped-" + genome + ext
cl = [
"bowtie",
"--solexa1.3-quals",
"--un",
out_file,
sam_ref,
"-1",
data["fastq1"],
"-2",
data["fastq2"],
"/dev/null",
]
log.info("Running %s" % cl)
subprocess.check_call(cl)
except KeyError:
log.error("Not removing genomes, directive filter_out_genomes undefined in run_info.yaml")
pass
def main(config_file, fc_dir, project_dir, run_info_yaml=None, fc_alias=None, project_desc=None, lanes=None):
if project_desc is None and lanes is None:
log.error("No project description or lanes provided: cannot deliver files without this information")
sys.exit()
config = load_config(config_file)
## Set log file in project output directory
config.update(log_dir=os.path.join(project_dir, "log"))
log_handler = create_log_handler(config, log.name)
fc_dir = os.path.normpath(fc_dir)
fc_name, fc_date, run_info = get_run_info(fc_dir, config, run_info_yaml)
with log_handler.applicationbound():
run_info = prune_run_info_by_description(run_info['details'], project_desc, lanes)
if len(run_info) == 0:
log.error("No lanes found with matching description %s: please check your flowcell run information" % project_desc)
sys.exit()
dirs = dict(fc_dir=fc_dir, project_dir=project_dir)
fc_name, fc_date = get_flowcell_id(run_info, dirs['fc_dir'])
config.update(fc_name = fc_name, fc_date = fc_date)
config.update(fc_alias = "%s_%s" % (fc_date, fc_name) if not fc_alias else fc_alias)
dirs.update(fc_delivery_dir = os.path.join(dirs['project_dir'], options.data_prefix, config['fc_alias'] ))
dirs.update(data_delivery_dir = os.path.join(dirs['project_dir'], options.data_prefix, "%s_%s" %(fc_date, fc_name) ))
with log_handler.applicationbound():
config = _make_delivery_directory(dirs, config)
_save_run_info(run_info, dirs['fc_delivery_dir'], run_exit=options.only_run_info)
run_main(run_info, config, dirs)
