def _prepare_input(job_context: Dict) -> Dict:
start_time = log_state("prepare input", job_context["job"].id)
job_context["primary_organism"] = max(
job_context["samples"],
key=lambda organism: len(job_context["samples"][organism]))
job_context["all_organisms"] = job_context["samples"].keys()
all_samples = list(itertools.chain(*job_context["samples"].values()))
job_context["samples"] = {job_context["primary_organism"]: all_samples}
# We'll store here all sample accession codes that didn't make it into the compendia
# with the reason why not.
job_context["filtered_samples"] = {}
job_context = smashing_utils.prepare_files(job_context)
# Compendia jobs only run for one organism, so we know the only
# key will be the organism name, unless of course we've already failed.
if job_context["job"].success is not False:
job_context["organism_name"] = job_context["group_by_keys"][0]
# TEMPORARY for iterating on compendia more quickly. Rather
# than downloading the data from S3 each run we're just gonna
# use the same directory every job.
job_context["old_work_dir"] = job_context["work_dir"]
job_context[
"work_dir"] = SMASHING_DIR + job_context["organism_name"] + "/"
if not os.path.exists(job_context["work_dir"]):
os.makedirs(job_context["work_dir"])
log_state("prepare input done", job_context["job"].id, start_time)
return job_context
def _prepare_input(job_context: Dict) -> Dict:
# We're going to use the smasher outside of the smasher.
# I'm not crazy about this yet. Maybe refactor later,
# but I need the data now.
job_context = smashing_utils.prepare_files(job_context)
# work_dir is already created by smasher._prepare_files
outfile_base = job_context["work_dir"] + str(time.time()).split(".")[0]
job_context["target_file"] = outfile_base + "_target.tsv"
return job_context