Merge branch 'master' of github.com:pystatgen/sgkit into regenie_wgr

Author: eric-czech

sgkit/api.py

@@ -59,25 +59,25 @@ def create_genotype_call_dataset(
     check_array_like(sample_id, kind="U", ndim=1)
     check_array_like(call_genotype, kind="i", ndim=3)
     data_vars: Dict[Hashable, Any] = {
-        "variant/contig": ([DIM_VARIANT], variant_contig),
-        "variant/position": ([DIM_VARIANT], variant_position),
-        "variant/alleles": ([DIM_VARIANT, DIM_ALLELE], variant_alleles),
-        "sample/id": ([DIM_SAMPLE], sample_id),
-        "call/genotype": ([DIM_VARIANT, DIM_SAMPLE, DIM_PLOIDY], call_genotype),
-        "call/genotype_mask": (
+        "variant_contig": ([DIM_VARIANT], variant_contig),
+        "variant_position": ([DIM_VARIANT], variant_position),
+        "variant_allele": ([DIM_VARIANT, DIM_ALLELE], variant_alleles),
+        "sample_id": ([DIM_SAMPLE], sample_id),
+        "call_genotype": ([DIM_VARIANT, DIM_SAMPLE, DIM_PLOIDY], call_genotype),
+        "call_genotype_mask": (
             [DIM_VARIANT, DIM_SAMPLE, DIM_PLOIDY],
             call_genotype < 0,
         ),
     }
     if call_genotype_phased is not None:
         check_array_like(call_genotype_phased, kind="b", ndim=2)
-        data_vars["call/genotype_phased"] = (
+        data_vars["call_genotype_phased"] = (
             [DIM_VARIANT, DIM_SAMPLE],
             call_genotype_phased,
         )
     if variant_id is not None:
         check_array_like(variant_id, kind="U", ndim=1)
-        data_vars["variant/id"] = ([DIM_VARIANT], variant_id)
+        data_vars["variant_id"] = ([DIM_VARIANT], variant_id)
     attrs: Dict[Hashable, Any] = {"contigs": variant_contig_names}
     return xr.Dataset(data_vars=data_vars, attrs=attrs)
 
@@ -124,15 +124,15 @@ def create_genotype_dosage_dataset(
     check_array_like(sample_id, kind="U", ndim=1)
     check_array_like(call_dosage, kind="f", ndim=2)
     data_vars: Dict[Hashable, Any] = {
-        "variant/contig": ([DIM_VARIANT], variant_contig),
-        "variant/position": ([DIM_VARIANT], variant_position),
-        "variant/alleles": ([DIM_VARIANT, DIM_ALLELE], variant_alleles),
-        "sample/id": ([DIM_SAMPLE], sample_id),
-        "call/dosage": ([DIM_VARIANT, DIM_SAMPLE], call_dosage),
-        "call/dosage_mask": ([DIM_VARIANT, DIM_SAMPLE], np.isnan(call_dosage),),
+        "variant_contig": ([DIM_VARIANT], variant_contig),
+        "variant_position": ([DIM_VARIANT], variant_position),
+        "variant_allele": ([DIM_VARIANT, DIM_ALLELE], variant_alleles),
+        "sample_id": ([DIM_SAMPLE], sample_id),
+        "call_dosage": ([DIM_VARIANT, DIM_SAMPLE], call_dosage),
+        "call_dosage_mask": ([DIM_VARIANT, DIM_SAMPLE], np.isnan(call_dosage),),
     }
     if variant_id is not None:
         check_array_like(variant_id, kind="U", ndim=1)
-        data_vars["variant/id"] = ([DIM_VARIANT], variant_id)
+        data_vars["variant_id"] = ([DIM_VARIANT], variant_id)
     attrs: Dict[Hashable, Any] = {"contigs": variant_contig_names}
     return xr.Dataset(data_vars=data_vars, attrs=attrs)

sgkit/stats/aggregation.py

@@ -15,7 +15,7 @@ def count_alleles(ds: Dataset) -> DataArray:
 
     Returns
     -------
-    variant/allele_count : DataArray
+    variant_allele_count : DataArray
         Allele counts with shape (variants, alleles) and values
         corresponding to the number of non-missing occurrences
         of each allele.
@@ -26,7 +26,7 @@ def count_alleles(ds: Dataset) -> DataArray:
     >>> import sgkit as sg
     >>> from sgkit.testing import simulate_genotype_call_dataset
     >>> ds = simulate_genotype_call_dataset(n_variant=4, n_sample=2, seed=1)
-    >>> ds['call/genotype'].to_series().unstack().astype(str).apply('/'.join, axis=1).unstack() # doctest: +NORMALIZE_WHITESPACE
+    >>> ds['call_genotype'].to_series().unstack().astype(str).apply('/'.join, axis=1).unstack() # doctest: +NORMALIZE_WHITESPACE
     samples 0   1
     variants
     0       1/0	1/0
@@ -42,8 +42,8 @@ def count_alleles(ds: Dataset) -> DataArray:
     """
     # Count each allele index individually as a 1D vector and
     # restack into new alleles dimension with same order
-    G = ds["call/genotype"].stack(calls=("samples", "ploidy"))
-    M = ds["call/genotype_mask"].stack(calls=("samples", "ploidy"))
+    G = ds["call_genotype"].stack(calls=("samples", "ploidy"))
+    M = ds["call_genotype_mask"].stack(calls=("samples", "ploidy"))
     n_variant, n_allele = G.shape[0], ds.dims["alleles"]
     max_allele = n_allele + 1
 
@@ -68,4 +68,4 @@ def count_alleles(ds: Dataset) -> DataArray:
     AC = CTS.sum(axis=0)[:, :n_allele]
     assert AC.shape == (n_variant, n_allele)
 
-    return DataArray(data=AC, dims=("variants", "alleles"), name="variant/allele_count")
+    return DataArray(data=AC, dims=("variants", "alleles"), name="variant_allele_count")

sgkit/stats/association.py

@@ -103,7 +103,7 @@ def _get_loop_covariates(ds: Dataset, dosage: Optional[str] = None) -> Array:
     if dosage is None:
         # TODO: This should be (probably gwas-specific) allele
         # count with sex chromosome considerations
-        G = ds["call/genotype"].sum(dim="ploidy")  # pragma: no cover
+        G = ds["call_genotype"].sum(dim="ploidy")  # pragma: no cover
     else:
         G = ds[dosage]
     return da.asarray(G.data)
@@ -217,8 +217,8 @@ def gwas_linear_regression(
     res = linear_regression(G.T, X, Y)
     return xr.Dataset(
         {
-            "variant/beta": (("variants", "traits"), res.beta),
-            "variant/t_value": (("variants", "traits"), res.t_value),
-            "variant/p_value": (("variants", "traits"), res.p_value),
+            "variant_beta": (("variants", "traits"), res.beta),
+            "variant_t_value": (("variants", "traits"), res.t_value),
+            "variant_p_value": (("variants", "traits"), res.p_value),
         }
     )

sgkit/testing.py

@@ -37,7 +37,7 @@ def simulate_genotype_call_dataset(
         Number of alleles to simulate
     n_contig : int, optional
         Number of contigs to partition variants with,
-        controlling values in `variant/contig`. Values
+        controlling values in `variant_contig`. Values
         will all be 0 by default with `n_contig` == 1.
     seed : int, optional
         Seed for random number generation

sgkit/tests/test_aggregation.py

@@ -14,9 +14,9 @@ def get_dataset(calls: ArrayLike, **kwargs: Any) -> Dataset:
     ds = simulate_genotype_call_dataset(
         n_variant=calls.shape[0], n_sample=calls.shape[1], **kwargs
     )
-    dims = ds["call/genotype"].dims
-    ds["call/genotype"] = xr.DataArray(calls, dims=dims)
-    ds["call/genotype_mask"] = xr.DataArray(calls < 0, dims=dims)
+    dims = ds["call_genotype"].dims
+    ds["call_genotype"] = xr.DataArray(calls, dims=dims)
+    ds["call_genotype_mask"] = xr.DataArray(calls < 0, dims=dims)
     return ds
 
 

sgkit/tests/test_api.py

@@ -41,14 +41,14 @@ def test_create_genotype_call_dataset():
     assert DIM_ALLELE in ds.dims
 
     assert ds.attrs["contigs"] == variant_contig_names
-    assert_array_equal(ds["variant/contig"], variant_contig)
-    assert_array_equal(ds["variant/position"], variant_position)
-    assert_array_equal(ds["variant/alleles"], variant_alleles)
-    assert_array_equal(ds["variant/id"], variant_id)
-    assert_array_equal(ds["sample/id"], sample_id)
-    assert_array_equal(ds["call/genotype"], call_genotype)
-    assert_array_equal(ds["call/genotype_mask"], call_genotype < 0)
-    assert_array_equal(ds["call/genotype_phased"], call_genotype_phased)
+    assert_array_equal(ds["variant_contig"], variant_contig)
+    assert_array_equal(ds["variant_position"], variant_position)
+    assert_array_equal(ds["variant_allele"], variant_alleles)
+    assert_array_equal(ds["variant_id"], variant_id)
+    assert_array_equal(ds["sample_id"], sample_id)
+    assert_array_equal(ds["call_genotype"], call_genotype)
+    assert_array_equal(ds["call_genotype_mask"], call_genotype < 0)
+    assert_array_equal(ds["call_genotype_phased"], call_genotype_phased)
 
 
 def test_create_genotype_dosage_dataset():
@@ -73,10 +73,10 @@ def test_create_genotype_dosage_dataset():
     assert DIM_SAMPLE in ds.dims
 
     assert ds.attrs["contigs"] == variant_contig_names
-    assert_array_equal(ds["variant/contig"], variant_contig)
-    assert_array_equal(ds["variant/position"], variant_position)
-    assert_array_equal(ds["variant/alleles"], variant_alleles)
-    assert_array_equal(ds["variant/id"], variant_id)
-    assert_array_equal(ds["sample/id"], sample_id)
-    assert_array_equal(ds["call/dosage"], call_dosage)
-    assert_array_equal(ds["call/dosage_mask"], np.isnan(call_dosage))
+    assert_array_equal(ds["variant_contig"], variant_contig)
+    assert_array_equal(ds["variant_position"], variant_position)
+    assert_array_equal(ds["variant_allele"], variant_alleles)
+    assert_array_equal(ds["variant_id"], variant_id)
+    assert_array_equal(ds["sample_id"], sample_id)
+    assert_array_equal(ds["call_dosage"], call_dosage)
+    assert_array_equal(ds["call_dosage_mask"], np.isnan(call_dosage))

sgkit/tests/test_association.py

@@ -129,7 +129,7 @@ def _get_statistics(
         res = _sm_statistics(ds, i, add_intercept)
         df_pred.append(
             dsr.to_dataframe()  # type: ignore[no-untyped-call]
-            .rename(columns=lambda c: c.replace("variant/", ""))
+            .rename(columns=lambda c: c.replace("variant_", ""))
             .iloc[i]
             .to_dict()
         )