def caseinfo_to_case(caseinfo: dict, sample_path: utils.URLPath) -> "Case":
# required keys
assert_in_dict(CASE_REQUIRED_FIELDS, caseinfo)
case_id = caseinfo["id"]
case_date = utils.str_to_date(caseinfo["date"])
samples = [
sample.sampleinfo_to_sample(sampleinfo, case_id, sample_path)
for sampleinfo in caseinfo["filepaths"]
]
# optional keys
infiltration = caseinfo.get("infiltration", 0.0)
infiltration = float(
infiltration.replace(",", ".") if isinstance(infiltration, str
) else infiltration)
assert infiltration <= 100.0 and infiltration >= 0.0, "Infiltration out of range 0-100"
group = caseinfo.get("cohort", "")
diagnosis = caseinfo.get("diagnosis", "")
sureness = caseinfo.get("sureness", "")
case = Case(
id=case_id,
date=case_date,
samples=samples,
infiltration=infiltration,
group=group,
diagnosis=diagnosis,
sureness=sureness,
)
return case
def sampleinfo_to_sample(sample_info: dict, case_id: str,
dataset_path: utils.URLPath) -> "Sample":
"""Create a tube sample from sample info dict."""
assert "fcs" in sample_info and "path" in sample_info[
"fcs"], "Path to sample_info is missing"
assert "date" in sample_info, "Date is missing"
path = utils.URLPath(sample_info["fcs"]["path"])
date = utils.str_to_date(sample_info["date"])
tube = str(sample_info.get("tube", "0"))
material = Material.from_str(sample_info.get("material", ""))
panel = sample_info.get("panel", "")
markers = sample_info["fcs"].get("markers", None)
count = int(sample_info["fcs"].get("event_count", 0)) or None
sample_id = f"{case_id}_t{tube}_{material.name}_{sample_info['date']}"
sample = FCSSample(id=sample_id,
case_id=case_id,
path=path,
dataset_path=dataset_path,
date=date,
tube=tube,
material=material,
panel=panel,
markers=markers,
count=count)
return sample
def do_ls(self, arg):
"List members of groups or cases."
if arg == "":
group_counts = self.data.group_count
for name, count in group_counts.items():
print(f"{name}: {count} cases")
else:
queries, save = tokenize(arg)
cases = list(self.data.data)
for query in queries:
if query["type"] == "g":
cases = [c for c in cases if c.group == query["query"]]
elif query["type"] == "ig":
cases = [
c for c in cases
if c.infiltration > float(query["query"])
]
elif query["type"] == "il":
cases = [
c for c in cases
if c.infiltration < float(query["query"])
]
elif query["type"] == "s":
cases = [
c for c in cases
if c.infiltration == int(query["query"])
]
elif query["type"] == "p":
cases = [
c for c in cases
if len(c.filepaths) == int(query["query"])
]
elif query["type"] == "dg":
date_min = utils.str_to_date(query["query"])
cases = [c for c in cases if c.date >= date_min]
elif query["type"] == "dl":
date_max = utils.str_to_date(query["query"])
cases = [c for c in cases if c.date <= date_max]
else:
print("Invalid type ", query["type"])
[info_case(c) for c in cases[:10]]
print(f"Total {len(cases)}")
if save:
print(f"Saving labels to {save}")
labels = [c.id for c in cases]
utils.save_json(labels, save)
def json_to_fcssample(samplejson: dict) -> "FCSSample":
samplejson["date"] = utils.str_to_date(samplejson["date"])
samplejson["path"] = utils.URLPath(samplejson["path"])
if samplejson["material"]:
samplejson["material"] = Material[samplejson["material"]]
else:
samplejson["material"] = None
return FCSSample(**samplejson)
def json_to_case(jscase: dict) -> "Case":
if jscase["used_material"]:
material = Material[jscase["used_material"]]
else:
material = None
jscase["used_material"] = material
jscase["date"] = utils.str_to_date(jscase["date"])
return Case(**jscase)
def json_to_somsample(samplejson: dict) -> "SOMSample":
samplejson["date"] = utils.str_to_date(samplejson["date"])
samplejson["path"] = utils.URLPath(samplejson["path"])
samplejson["dims"] = tuple(samplejson["dims"])
return SOMSample(**samplejson)
if labels and case.id not in labels:
reasons.append(f"labels")
if infiltration and case.group != "normal":
infiltration_min, infiltration_max = infiltration
if infiltration_min and case.infiltration < infiltration_min:
reasons.append("infiltration_min")
if infiltration_max and case.infiltration > infiltration_max:
reasons.append("infiltration_max")
if date:
date_min, date_max = date
if date_min:
if isinstance(date_min, str):
date_min = utils.str_to_date(date_min)
if case.date < date_min:
reasons.append("date_min")
if date_max:
if isinstance(date_max, str):
date_max = utils.str_to_date(date_max)
if case.date > date_max:
reasons.append("date_max")
if counts:
assert tubes, "Tubes required for counts"
if not any(case.get_tube(t, min_count=counts) for t in tubes):
reasons.append("counts")
if materials: