def test_include_tasks_exclude_task_none(self) -> None:
# We treat falsy values as unspecified, so the full list of task_names should be returned.
tasks = resolve_tasks()
assert tasks == TASKS
tasks = resolve_tasks(include_tasks="")
assert tasks == TASKS
tasks = resolve_tasks(include_tasks=[])
assert tasks == TASKS
def prepare(
filepath: Path = typer.Argument(
..., writable=True, help="Filepath to save prepared benchmark file."),
include_tasks: List[str] = typer.Option(
None,
help=
"A task name (or list of task names) to include in the prepared data."
),
exclude_tasks: List[str] = typer.Option(
None,
help=
"A task name (or list of task names) to exclude in the prepared data."
),
):
tasks = resolve_tasks(include_tasks, exclude_tasks)
benchmark = load_benchmark()
prepared_benchmark = {task: benchmark[task] for task in tasks}
task_specific_genes = list(
set([
gene for partition in prepared_benchmark.values()
for genes in partition.values() for gene in genes.keys()
])) # Find the subset of genes specific to the `tasks` specified
prepared_benchmark["inputs"] = {
gene: benchmark["inputs"][gene]
for gene in task_specific_genes
}
prepared_benchmark = orjson.dumps(prepared_benchmark,
option=orjson.OPT_INDENT_2)
filepath.write_bytes(prepared_benchmark)
def __init__(
self,
features: pd.DataFrame,
include_tasks: Optional[Union[str, List[str]]] = None,
exclude_tasks: Optional[Union[str, List[str]]] = None,
) -> None:
self._features = features
self._tasks = resolve_tasks(include_tasks, exclude_tasks)
self.results = {}
def test_value_error_include_and_exclude(self) -> None:
with pytest.raises(ValueError):
_ = resolve_tasks(include_tasks=["arbitrary"],
exclude_tasks=["arbitrary"])
def test_exclude_tasks(self) -> None:
task_name = list(TASKS)[0]
expected = TASKS - {task_name}
actual = resolve_tasks(exclude_tasks=task_name)
assert expected == actual
def test_value_error_invalid_task_names(self, tasks: List[str]) -> None:
with pytest.raises(ValueError):
_ = resolve_tasks(include_tasks=tasks)
with pytest.raises(ValueError):
_ = resolve_tasks(exclude_tasks=tasks)