add flake8 pre-commit

.flake8

@@ -1,20 +1,37 @@
 # Can't yet be moved to the pyproject.toml due to https://gitlab.com/pycqa/flake8/-/issues/428#note_251982786
 [flake8]
 max-line-length = 88
-# switched off since they conflict with black's standards
-ignore = F401,  # module imported but unused -> required for Scanpys API
-         W503,  # line break before a binary operator -> black does not adhere to PEP8
-         W504,  # line break occured after a binary operator -> black does not adhere to PEP8
-         E501,  # line too long -> we accept long comment lines; black gets rid of long code lines
-         E203,  # whitespace before : -> black does not adhere to PEP8
-         E231,  # missing whitespace after ,', ';', or ':' -> black does not adhere to PEP8
-         E402,  # module level import not at top of file -> required to circumvent circular imports for Scanpys API
-         E126,  # continuation line over-indented for hanging indent -> black does not adhere to PEP8
-         E266,  # E266 too many leading '#' for block comment -> Scanpy allows them for comments into sections
-         E731   # Do not assign a lambda expression, use a def -> Scanpy allows lambda expression assignments
+ignore = # module imported but unused -> required for Scanpys API
+         F401,
+         # line break before a binary operator -> black does not adhere to PEP8
+         W503,
+         # line break occured after a binary operator -> black does not adhere to PEP8
+         W504,
+         # line too long -> we accept long comment lines; black gets rid of long code lines
+         E501,
+         # whitespace before : -> black does not adhere to PEP8
+         E203,
+         # missing whitespace after ,', ';', or ':' -> black does not adhere to PEP8
+         E231,
+         # module level import not at top of file -> required to circumvent circular imports for Scanpys API
+         E402,
+         # continuation line over-indented for hanging indent -> black does not adhere to PEP8
+         E126,
+         # E266 too many leading '#' for block comment -> Scanpy allows them for comments into sections
+         E262,
+         # inline comment should start with '# ' -> Scanpy allows them for specific explanations
+         E266,
+         # Do not assign a lambda expression, use a def -> Scanpy allows lambda expression assignments,
+         E731,
+         # allow I, O, l as variable names -> I is the identity matrix, i, j, k, l is reasonable indexing notation
+         E741
+ per-file-ignores =
+    # F811 Redefinition of unused name from line, does not play nice with pytest fixtures
+    tests/test*.py: F811
 exclude =
     .git,
     __pycache__,
     build,
-    docs/_build
-    dist
+    docs/_build,
+    dist,
+

.pre-commit-config.yaml

@@ -7,3 +7,8 @@ repos:
     rev: 3.8.4
     hooks:
     -   id: flake8
+-   repo: https://github.com/pre-commit/mirrors-autopep8
+    rev: v1.5.5
+    hooks:
+    -   id: autopep8
+        args: ["-i"]

docs/dev/code.rst

@@ -16,11 +16,10 @@ Code style
 ----------
 
 New code should follow
-`Black <https://black.readthedocs.io/en/stable/the_black_code_style.html>`__,
+`Black <https://black.readthedocs.io/en/stable/the_black_code_style.html>`__
 and
 `flake8 <https://flake8.pycqa.org>`__.
-We ignore a couple of flake8 checks which are documented in the .flake8 file in the root
-of this repository.
+We ignore a couple of flake8 checks which are documented in the .flake8 file in the root of this repository.
 To learn how to ignore checks per line please read
 `flake8 violations <https://flake8.pycqa.org/en/latest/user/violations.html>`__.
 Additionally, we use Scanpy’s

scanpy/_utils.py

@@ -554,7 +554,9 @@ def warn_with_traceback(message, category, filename, lineno, file=None, line=Non
     import traceback
 
     traceback.print_stack()
-    log = file if hasattr(file, 'write') else sys.stderr  # noqa: F841
+    log = (  # noqa: F841  # TODO Does this need fixing?
+        file if hasattr(file, 'write') else sys.stderr
+    )
     settings.write(warnings.formatwarning(message, category, filename, lineno, line))
 
 

scanpy/external/pl.py

@@ -334,7 +334,8 @@ def scrublet_score_distribution(
     """\
     Plot histogram of doublet scores for observed transcriptomes and simulated doublets.
 
-    The histogram for simulated doublets is useful for determining the correct doublet score threshold.
+    The histogram for simulated doublets is useful for determining the correct doublet
+    score threshold.
 
     Parameters
     ----------

scanpy/get/get.py

@@ -6,7 +6,6 @@
 from scipy.sparse import spmatrix
 
 from anndata import AnnData
-import warnings
 
 # --------------------------------------------------------------------------------
 # Plotting data helpers
@@ -96,7 +95,7 @@ def rank_genes_groups_df(
 def _check_indices(
     dim_df: pd.DataFrame,
     alt_index: pd.Index,
-    dim: "Literal['obs', 'var']",  # noqa: F821
+    dim: "Literal['obs', 'var']",  # noqa: F821  # TODO Does this need fixing?
     keys: List[str],
     alias_index: Optional[pd.Index] = None,
     use_raw: bool = False,
@@ -176,7 +175,7 @@ def _get_array_values(
     X,
     dim_names: pd.Index,
     keys: List[str],
-    axis: "Literal[0, 1]",  # noqa: F821
+    axis: "Literal[0, 1]",  # noqa: F821  # TODO Does this need fixing?
     backed: bool,
 ):
     # TODO: This should be made easier on the anndata side

scanpy/neighbors/__init__.py

@@ -860,7 +860,7 @@ def _compute_connectivities_diffmap(self, density_normalize=True):
             # make the weight matrix sparse
             if not self.knn:
                 mask = W > 1e-14
-                W[mask == False] = 0  # noqa: E712
+                W[~mask] = 0
             else:
                 # restrict number of neighbors to ~k
                 # build a symmetric mask
@@ -872,7 +872,7 @@ def _compute_connectivities_diffmap(self, density_normalize=True):
                             W[j, i] = W[i, j]
                             mask[j, i] = True
                 # set all entries that are not nearest neighbors to zero
-                W[mask == False] = 0  # noqa: E712
+                W[~mask] = 0
         else:
             W = (
                 Dsq.copy()
@@ -1023,14 +1023,14 @@ def _get_dpt_row(self, i):
             mask = self._connected_components[1] == label
         row = sum(
             (
-                self.eigen_values[k]
-                / (1 - self.eigen_values[k])
-                * (self.eigen_basis[i, k] - self.eigen_basis[:, k])
+                self.eigen_values[j]
+                / (1 - self.eigen_values[j])
+                * (self.eigen_basis[i, j] - self.eigen_basis[:, j])
             )
             ** 2
             # account for float32 precision
-            for k in range(0, self.eigen_values.size)
-            if self.eigen_values[k] < 0.9994
+            for j in range(0, self.eigen_values.size)
+            if self.eigen_values[j] < 0.9994
         )
         # thanks to Marius Lange for pointing Alex to this:
         # we will likely remove the contributions from the stationary state below when making

scanpy/plotting/_tools/__init__.py

@@ -62,7 +62,7 @@ def pca_overview(adata: AnnData, **params):
     show = params['show'] if 'show' in params else None
     if 'show' in params:
         del params['show']
-    scatterplots.pca(adata, **params, show=False)  # noqa: F821
+    pca(adata, **params, show=False)
     pca_loadings(adata, show=False)
     pca_variance_ratio(adata, show=show)
 
@@ -165,6 +165,7 @@ def dpt_timeseries(
 ):
     """\
     Heatmap of pseudotime series.
+
     Parameters
     ----------
     as_heatmap
@@ -364,7 +365,7 @@ def _fig_show_save_or_axes(plot_obj, return_fig, show, save):
         plot_obj.make_figure()
         savefig_or_show(plot_obj.DEFAULT_SAVE_PREFIX, show=show, save=save)
         show = settings.autoshow if show is None else show
-        if not show:
+        if show == False:  # noqa: E712  # TODO Does this need fixing? -> is False
             return plot_obj.get_axes()
 
 
@@ -966,7 +967,7 @@ def rank_genes_groups_violin(
         )
         savefig_or_show(writekey, show=show, save=save)
         axs.append(_ax)
-    if not show:
+    if show == False:  # noqa: E712  TODO Does this need fixing?
         return axs
 
 

scanpy/plotting/_tools/paga.py

@@ -148,7 +148,7 @@ def paga_compare(
     if suptitle is not None:
         pl.suptitle(suptitle)
     _utils.savefig_or_show('paga_compare', show=show, save=save)
-    if not show:
+    if show == False:  # noqa: E712  # TODO Does this need fixing?
         return axs
 
 

scanpy/plotting/_utils.py

@@ -620,7 +620,9 @@ def setup_axes(
     figure_width = width_without_offsets + left_offset + right_offset
     draw_region_width_frac = draw_region_width / figure_width
     left_offset_frac = left_offset / figure_width
-    right_offset_frac = 1 - (len(panels) - 1) * left_offset_frac  # noqa: F841
+    right_offset_frac = (  # noqa: F841  # TODO Does this need fixing?
+        1 - (len(panels) - 1) * left_offset_frac
+    )
 
     if ax is None:
         pl.figure(
@@ -708,9 +710,7 @@ def scatter_base(
     )
     for icolor, color in enumerate(colors):
         ax = axs[icolor]
-        left = panel_pos[2][2 * icolor]  # noqa: F841
         bottom = panel_pos[0][0]
-        width = draw_region_width / figure_width  # noqa: F841
         height = panel_pos[1][0] - bottom
         Y_sort = Y
         if not is_color_like(color) and sort_order:
@@ -940,8 +940,8 @@ def make_pos(pos, node=root, currentLevel=0, parent=None, vert_loc=0):
     if levels is None:
         levels = make_levels({})
     else:
-        levels = {level: {TOTAL: levels[level], CURRENT: 0} for level in levels}
-    vert_gap = height / (max([level for level in levels]) + 1)
+        levels = {k: {TOTAL: v, CURRENT: 0} for k, v in levels.items()}
+    vert_gap = height / (max(levels.keys()) + 1)
     return make_pos({})
 
 

scanpy/preprocessing/_deprecated/highly_variable_genes.py

@@ -170,10 +170,9 @@ def filter_genes_dispersion(
         disp_mean_bin[one_gene_per_bin] = 0
         # actually do the normalization
         df['dispersion_norm'] = (
+            # use values here as index differs
             df['dispersion'].values
-            - disp_mean_bin[  # use values here as index differs
-                df['mean_bin'].values
-            ].values
+            - disp_mean_bin[df['mean_bin'].values].values
         ) / disp_std_bin[df['mean_bin'].values].values
     elif flavor == 'cell_ranger':
         from statsmodels import robust

scanpy/preprocessing/_highly_variable_genes.py

@@ -236,10 +236,8 @@ def _highly_variable_genes_single_batch(
         disp_mean_bin[one_gene_per_bin.values] = 0
         # actually do the normalization
         df['dispersions_norm'] = (
-            df['dispersions'].values
-            - disp_mean_bin[
-                df['mean_bin'].values
-            ].values  # use values here as index differs
+            df['dispersions'].values  # use values here as index differs
+            - disp_mean_bin[df['mean_bin'].values].values
         ) / disp_std_bin[df['mean_bin'].values].values
     elif flavor == 'cell_ranger':
         from statsmodels import robust

scanpy/preprocessing/_simple.py

@@ -752,9 +752,11 @@ def scale(
     annotated with `'mean'` and `'std'` in `adata.var`.
     """
     _check_array_function_arguments(layer=layer, obsm=obsm)
-    return scale_array(
-        data, zero_center=zero_center, max_value=max_value, copy=copy  # noqa: F821
-    )
+    if layer is not None:
+        raise ValueError(f"`layer` argument inappropriate for value of type {type(X)}")
+    if obsm is not None:
+        raise ValueError(f"`obsm` argument inappropriate for value of type {type(X)}")
+    return scale_array(X, zero_center=zero_center, max_value=max_value, copy=copy)
 
 
 @scale.register(np.ndarray)

scanpy/tests/test_pca.py

@@ -38,7 +38,7 @@
 )
 
 
-def test_pca_transform(array_type):  # noqa: F811
+def test_pca_transform(array_type):
     A = array_type(A_list).astype('float32')
     A_pca_abs = np.abs(A_pca)
     A_svd_abs = np.abs(A_svd)
@@ -101,7 +101,7 @@ def test_pca_sparse(pbmc3k_normalized):
 
 # This will take a while to run, but irreproducibility may
 # not show up for float32 unless the matrix is large enough
-def test_pca_reproducible(pbmc3k_normalized, array_type, float_dtype):  # noqa: F811
+def test_pca_reproducible(pbmc3k_normalized, array_type, float_dtype):
     pbmc = pbmc3k_normalized
     pbmc.X = array_type(pbmc.X)
 

scanpy/tools/_dpt.py

@@ -342,7 +342,9 @@ def check_adjacency(self):
         for n_edges in range(1, np.max(n_edges_per_seg) + 1):
             for iseg in range(self.segs_adjacency.shape[0]):
                 if n_edges_per_seg[iseg] == n_edges:
-                    _ = self.segs_adjacency[iseg].todense().A1
+                    neighbor_segs = (  # noqa: F841  TODO Evaluate whether to assign the variable or not
+                        self.segs_adjacency[iseg].todense().A1
+                    )
                     closest_points_other_segs = [
                         seg[np.argmin(self.distances_dpt[self.segs_tips[iseg][0], seg])]
                         for seg in self.segs
@@ -593,7 +595,9 @@ def detect_branching(
                 for iseg, seg_connects in enumerate(ssegs_connects)
                 if iseg != trunk
             ]
-            _ = segs_connects[iseg]
+            prev_connecting_points = segs_connects[  # noqa: F841  TODO Evaluate whether to assign the variable or not
+                iseg
+            ]
             for jseg_cnt, jseg in enumerate(prev_connecting_segments):
                 iseg_cnt = 0
                 for iseg_new, seg_new in enumerate(ssegs):
@@ -899,9 +903,9 @@ def _detect_branching_single_haghverdi16(self, Dseg, tips):
         # permutations of tip cells
         ps = [
             [0, 1, 2],  # start by computing distances from the first tip
-            [1, 2, 0],  # -"-                       second tip
+            [1, 2, 0],  #             -"-                       second tip
             [2, 0, 1],
-        ]  # -"-                       third tip
+        ]  #             -"-                       third tip
         for i, p in enumerate(ps):
             ssegs.append(self.__detect_branching_haghverdi16(Dseg, tips[p]))
         return ssegs

scanpy/tools/_rank_genes_groups.py

@@ -37,7 +37,6 @@ def _ranks(X, mask=None, mask_rest=None):
     if issparse(X):
         merge = lambda tpl: vstack(tpl).toarray()
         adapt = lambda X: X.toarray()
-
     else:
         merge = np.vstack
         adapt = lambda X: X
@@ -49,7 +48,6 @@ def _ranks(X, mask=None, mask_rest=None):
         get_chunk = lambda X, left, right: merge(
             (X[mask, left:right], X[mask_rest, left:right])
         )
-
     else:
         n_cells = X.shape[0]
         get_chunk = lambda X, left, right: adapt(X[:, left:right])
@@ -171,7 +169,6 @@ def _basic_stats(self):
 
         if issparse(self.X):
             get_nonzeros = lambda X: X.getnnz(axis=0)
-
         else:
             get_nonzeros = lambda X: np.count_nonzero(X, axis=0)
 
@@ -744,7 +741,6 @@ def filter_rank_genes_groups(
 
         if 'log1p' in adata.uns_keys() and adata.uns['log1p']['base'] is not None:
             expm1_func = lambda x: np.expm1(x * np.log(adata.uns['log1p']['base']))
-
         else:
             expm1_func = np.expm1