def test_annos_dependening_none(self):
class A(Annotator):
cache_name = "hello"
columns = ["aa"]
def calc(self, df):
return pd.DataFrame({self.columns[0]: "a"}, index=df.index)
class B(Annotator):
cache_name = "hello2"
columns = ["ab"]
def calc(self, df):
return df["aa"] + "b"
def dep_annos(self):
return [None, A(), None]
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
a += B()
a.annotate()
assert "ab" in a.df.columns
assert "aa" in a.df.columns
assert (a.df["ab"] == (a.df["aa"] + "b")).all()
def test_annos_same_column_different_anno(self):
count = [0]
class CountingConstant(Annotator):
def __init__(self, column_name, value):
count[0] += 1
self.columns = [column_name]
self.value = value
def calc(self, df):
return pd.DataFrame({self.columns[0]: self.value}, index=df.index)
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
c = CountingConstant("hello", "c")
a += c
a.annotate()
assert "hello" in a.df.columns
assert count[0] == 1
c = CountingConstant("hello2", "c")
a += c
a.annotate()
assert "hello2" in a.df.columns
assert count[0] == 2
d = CountingConstant("hello2", "d")
assert c is not d
with pytest.raises(ValueError):
a += d
def test_annotator(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
a += Constant("column", "value")
a.annotate()
assert "column" in a.df.columns
assert (a.df["column"] == "value").all()
def test_annos_dependening(self):
class A(Annotator):
cache_name = "hello"
columns = ["aa"]
def calc(self, df):
return pd.DataFrame({self.columns[0]: "a"}, index=df.index)
class B(Annotator):
cache_name = "hello2"
columns = ["ab"]
def calc(self, df):
return df["aa"] + "b"
def dep_annos(self):
return [A()]
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
a += B()
ppg.JobGeneratingJob("shu", lambda: 55).depends_on(a.annotate())
ppg.run_pipegraph()
assert "ab" in a.df.columns
assert "aa" in a.df.columns
assert (a.df["ab"] == (a.df["aa"] + "b")).all()
def test_annotator_basic(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
a += Constant("aa", "aa")
force_load(a.annotate())
ppg.run_pipegraph()
assert (a.df["aa"] == "aa").all()
def test_DynamicColumNames(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
class Dynamic(Annotator):
@property
def columns(self):
return ["a"]
def calc(self, df):
return pd.DataFrame({"a": ["x", "y"]})
a += Dynamic()
a.annotate()
assert_frame_equal(
a.df, pd.DataFrame({"A": [1, 2], "B": ["c", "d"], "a": ["x", "y"]})
)
def test_missing_external_genome(self):
g = DelayedDataFrame("ex",
pd.DataFrame({"gene_stable_id": ["a", "c", "b"]}))
anno = genes.annotators.Description()
g += anno
force_load(g.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "ddf had no .genome and no genome was passed to Description" in str(
g.anno_jobs[anno.get_cache_name()].lfg.exception)
def test_annos_added_only_once(self):
count = [0]
class CountingConstant(Annotator):
def __init__(self, column_name, value):
count[0] += 1
self.columns = [column_name]
self.value = value
def calc(self, df):
return pd.DataFrame({self.columns[0]: self.value}, index=df.index)
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
c = CountingConstant("hello", "c")
a += c
a.annotate()
assert "hello" in a.df.columns
assert count[0] == 1
a += c # this get's ignored
def test_annotator_coliding_with_non_anno_column(self):
a = DelayedDataFrame(
"shu",
lambda: pd.DataFrame(
{"A": [1, 2, 3], "B": ["a", "b", "c"], "idx": ["x", "y", "z"]}
).set_index("idx"),
)
a += Constant("A", "aa")
lj = a.anno_jobs["A"]
ppg.JobGeneratingJob("shu", lambda: 55).depends_on(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "were already present" in str(lj().exception)
def test_anno_returing_right_length_but_wrong_start_range_index(self):
a = DelayedDataFrame("shu", lambda: pd.DataFrame({"A": [1, 2, 3]}))
class BadAnno(Annotator):
columns = ["X"]
def calc(self, df):
return pd.Series(["a", "b", "c"], index=pd.RangeIndex(5, 5 + 3))
a += BadAnno()
force_load(a.annotate())
lj = a.anno_jobs["X"]
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "Index mismatch" in str(lj().exception)
def test_anno_not_returning_enough_rows_and_no_index_range_index_on_df(self):
class BrokenAnno(Annotator):
columns = ["X"]
def calc(self, df):
return pd.DataFrame({"X": [1]})
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2, 3], "B": ["a", "b", "c"]})
)
a += BrokenAnno()
lj = a.anno_jobs["X"]
ppg.JobGeneratingJob("shu", lambda: 55).depends_on(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "Length and index mismatch " in str(lj().exception)
def test_anno_returning_series(self):
a = DelayedDataFrame(
"shu",
lambda: pd.DataFrame(
{"A": [1, 2, 3], "B": ["a", "b", "c"], "idx": ["x", "y", "z"]}
).set_index("idx"),
)
class SeriesAnno(Annotator):
columns = ["C"]
def calc(self, df):
return pd.Series(list(range(len(df))))
a += SeriesAnno()
ppg.JobGeneratingJob("shu", lambda: 55).depends_on(a.annotate())
ppg.run_pipegraph()
assert (a.df["C"] == [0, 1, 2]).all()
def test_annotator_raising(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
class RaiseAnno(Annotator):
columns = ["aa"]
cache_name = "empty"
def calc(self, df):
raise ValueError("hello")
anno1 = RaiseAnno()
a += anno1
force_load(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
anno_job = a.anno_jobs[RaiseAnno().get_cache_name()]
assert "hello" in str(anno_job.lfg.exception)
def test_DynamicColumNames(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
class Dynamic(Annotator):
@property
def columns(self):
return ["a"]
def calc(self, df):
return pd.DataFrame({"a": ["x", "y"]})
a += Dynamic()
a.anno_jobs[Dynamic().get_cache_name()]
force_load(a.annotate())
ppg.run_pipegraph()
assert_frame_equal(
a.df, pd.DataFrame({"A": [1, 2], "B": ["c", "d"], "a": ["x", "y"]})
)
def test_annotator_missing_columns(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
class MissingColumnNames(Annotator):
cache_name = "MissingColumnNames"
def calc(self, df):
return pd.DataFrame({})
def __repr__(self):
return "MissingColumnNames()"
a += MissingColumnNames()
lg = a.anno_jobs["MissingColumnNames"]
force_load(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "AttributeError" in repr(lg().lfg.exception)
def test_anno_returning_string(self):
a = DelayedDataFrame(
"shu",
lambda: pd.DataFrame(
{"A": [1, 2, 3], "B": ["a", "b", "c"], "idx": ["x", "y", "z"]}
).set_index("idx"),
)
class SeriesAnno(Annotator):
columns = ["C", "D"]
def calc(self, df):
return "abc"
a += SeriesAnno()
lj = a.anno_jobs["C"]
ppg.JobGeneratingJob("shu", lambda: 55).depends_on(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "result was no dataframe" in str(lj().lfg.exception)
def test_lying_about_columns(self):
a = DelayedDataFrame(
"shu",
lambda: pd.DataFrame(
{"A": [1, 2, 3], "B": ["a", "b", "c"], "idx": ["x", "y", "z"]}
).set_index("idx"),
)
class SeriesAnno(Annotator):
columns = ["C"]
def calc(self, df):
return pd.DataFrame({"D": [0, 1, 2]})
a += SeriesAnno()
lj = a.anno_jobs["C"]
ppg.JobGeneratingJob("shu", lambda: 55).depends_on(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "declared different " in str(lj().exception)
def test_annotator_columns_not_list(self):
class BrokenAnno(Annotator):
def __init__(
self,
):
self.columns = "shu"
def calc(self, df):
return pd.DataFrame(
{self.columns[0]: ["%s%i" % (self.columns[0], len(df))] * len(df)}
)
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
a += BrokenAnno()
lg = a.anno_jobs[BrokenAnno().get_cache_name()]
force_load(a.annotate())
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert "list" in str(lg().lfg.exception)
def test_annotator_empty_columns(self):
a = DelayedDataFrame(
"shu", lambda: pd.DataFrame({"A": [1, 2], "B": ["c", "d"]})
)
class EmptyColumnNames(Annotator):
columns = []
cache_name = "empty"
def calc(self, df):
return pd.DataFrame({"shu": [1, 2]})
def __repr__(self):
return "EmptyColumNames()"
a += EmptyColumnNames()
force_load(a.annotate())
anno_job_cb = a.anno_jobs[EmptyColumnNames().get_cache_name()]
with pytest.raises(ppg.RuntimeError):
ppg.run_pipegraph()
assert anno_job_cb() is anno_job_cb()
assert "anno.columns was empty" in repr(anno_job_cb().exception)
def test_external_genome(self):
genome = MockGenome(
pd.DataFrame({
"stable_id": ["a", "b", "c"],
"chr": "1",
"tss": [0, 100, 1000],
"tes": [10, 101, 1010],
}),
df_genes_meta=pd.DataFrame({
"gene_stable_id": ["a", "b", "c"],
"description": ["hello", "world", "!"],
}).set_index("gene_stable_id"),
)
g = DelayedDataFrame("ex",
pd.DataFrame({"gene_stable_id": ["a", "c", "b"]}))
anno = genes.annotators.Description(genome)
g += anno
force_load(g.annotate())
ppg.run_pipegraph()
assert "description" in g.df.columns
assert (g.df.sort_values("gene_stable_id")["description"] == [
"hello", "world", "!"
]).all()