Pin cyvcf2

.pre-commit-config.yaml

@@ -12,7 +12,7 @@ repos:
     hooks:
       - id: isort
   - repo: https://github.com/python/black
-    rev: 21.10b0
+    rev: 22.3.0
     hooks:
       - id: black
         language_version: python3

requirements-dev.txt

@@ -14,7 +14,7 @@ partd
 bed-reader
 rechunker == 0.3.3
 cbgen
-cyvcf2; platform_system != "Windows"
+cyvcf2 < 0.30.15; platform_system != "Windows"
 yarl
 matplotlib
 asv

requirements-doc.txt

@@ -2,9 +2,6 @@ myst_nb
 pydata-sphinx-theme
 sphinx>4,<=4.2.0
 sphinx_autodoc_typehints>=1.14.0
-# Work around problem with autodoc_typehints
-# https://github.com/theislab/scanpydoc/pull/32
-git+https://github.com/theislab/scanpydoc.git#egg=scanpydoc
 sphinx-book-theme
 scanpydoc
 ipython

setup.cfg

@@ -60,7 +60,7 @@ plink =
     bed-reader
 vcf =
     aiohttp
-    cyvcf2
+    cyvcf2 < 0.30.15
     requests
     yarl
 bgen =

sgkit/stats/aggregation.py

@@ -916,7 +916,7 @@ def individual_heterozygosity(
     # use nan denominator to avoid divide by zero with K - 1
     K2 = da.where(K > 1, K, np.nan)
     AF = AC / K2[..., None]
-    HI = (1 - da.sum(AF ** 2, axis=-1)) * (K / (K2 - 1))
+    HI = (1 - da.sum(AF**2, axis=-1)) * (K / (K2 - 1))
     new_ds = create_dataset(
         {variables.call_heterozygosity: (("variants", "samples"), HI)}
     )

sgkit/stats/association.py

@@ -78,15 +78,15 @@ def linear_regression(
     # from projection require no extra terms in variance
     # estimate for loop covariates (columns of G), which is
     # only true when an intercept is present.
-    XLPS = (XLP ** 2).sum(axis=0, keepdims=True).T
+    XLPS = (XLP**2).sum(axis=0, keepdims=True).T
     assert XLPS.shape == (n_loop_covar, 1)
     B = (XLP.T @ YP) / XLPS
     assert B.shape == (n_loop_covar, n_outcome)
 
     # Compute residuals for each loop covariate and outcome separately
     YR = YP[:, np.newaxis, :] - XLP[..., np.newaxis] * B[np.newaxis, ...]
     assert YR.shape == (n_obs, n_loop_covar, n_outcome)
-    RSS = (YR ** 2).sum(axis=0)
+    RSS = (YR**2).sum(axis=0)
     assert RSS.shape == (n_loop_covar, n_outcome)
     # Get t-statistics for coefficient estimates
     T = B / np.sqrt(RSS / dof / XLPS)
@@ -382,7 +382,7 @@ def regenie_loco_regression(
     Y -= Y.mean(axis=0)
     # Orthogonally project covariates out of phenotype matrix
     Y -= Q @ (Q.T @ Y)
-    Y_scale = da.sqrt(da.sum(Y ** 2, axis=0) / (Y.shape[0] - Q.shape[1]))
+    Y_scale = da.sqrt(da.sum(Y**2, axis=0) / (Y.shape[0] - Q.shape[1]))
     # Scale
     Y /= Y_scale[None, :]
 

sgkit/stats/ld.py

@@ -41,8 +41,8 @@ def rogers_huff_r_between(gn0: ArrayLike, gn1: ArrayLike) -> float:  # pragma: n
             n += 1
             m0 += x
             m1 += y
-            v0 += x ** 2
-            v1 += y ** 2
+            v0 += x**2
+            v1 += y**2
             cov += x * y
 
     # early out

sgkit/stats/popgen.py

@@ -591,11 +591,11 @@ def Tajimas_D(
     # calculate the denominator (standard deviation)
     a2 = (1 / (da.arange(1, n) ** 2)).sum()
     b1 = (n + 1) / (3 * (n - 1))
-    b2 = 2 * (n ** 2 + n + 3) / (9 * n * (n - 1))
+    b2 = 2 * (n**2 + n + 3) / (9 * n * (n - 1))
     c1 = b1 - (1 / a1)
-    c2 = b2 - ((n + 2) / (a1 * n)) + (a2 / (a1 ** 2))
+    c2 = b2 - ((n + 2) / (a1 * n)) + (a2 / (a1**2))
     e1 = c1 / a1
-    e2 = c2 / (a1 ** 2 + a2)
+    e2 = c2 / (a1**2 + a2)
     d_stdev = da.sqrt((e1 * S) + (e2 * S * (S - 1)))
 
     # Let IEEE decide the semantics of division by zero here. The return value
@@ -770,7 +770,7 @@ def _Garud_h(haplotypes: ArrayLike) -> ArrayLike:
     f = counts / n  # type: ignore[operator]
 
     # compute H1
-    h1 = np.sum(f ** 2)
+    h1 = np.sum(f**2)
 
     # compute H12
     h12 = np.sum(f[:2]) ** 2 + np.sum(f[2:] ** 2)  # type: ignore[index]