Merge pull request #323 from theislab/feature/display_qc_metrics

[FEATURE] Display QC metrics of var #239

docs/usage/api.rst

@@ -61,7 +61,7 @@ Quality control
 .. autosummary::
     :toctree: preprocessing
 
-    preprocessing.calculate_qc_metrics
+    preprocessing.qc_metrics
 
 Imputation
 ++++++++++
@@ -198,6 +198,14 @@ Generic
     plot.ranking
     plot.dendrogram
 
+Specific
+++++++++
+
+.. autosummary::
+    :toctree: plot
+
+    plot.qc_metrics
+
 Classes
 +++++++
 

docs/usage/plot/ehrapy.plot.qc_metrics.rst

@@ -0,0 +1,12 @@
+:github_url: ehrapy.plot.qc_metrics
+
+ehrapy.plot.qc\_metrics
+=======================
+
+.. autofunction:: ehrapy.plot.qc_metrics
+
+.. _sphx_glr_backref_ehrapy.plot.qc_metrics:
+
+.. minigallery:: ehrapy.plot.qc_metrics
+   :add-heading: Gallery
+   :heading-level: -

docs/usage/preprocessing/ehrapy.preprocessing.qc_metrics.rst

@@ -0,0 +1,12 @@
+:github_url: ehrapy.preprocessing.qc_metrics
+
+ehrapy.preprocessing.qc\_metrics
+================================
+
+.. autofunction:: ehrapy.preprocessing.qc_metrics
+
+.. _sphx_glr_backref_ehrapy.preprocessing.qc_metrics:
+
+.. minigallery:: ehrapy.preprocessing.qc_metrics
+   :add-heading: Gallery
+   :heading-level: -

ehrapy/plot/__init__.py

@@ -1,2 +1,3 @@
 from ehrapy.plot._plot_util import *  # noqa: E402,F403
 from ehrapy.plot._scanpy_pl_api import *  # noqa: E402,F403
+from ehrapy.plot.ehrapy_plot.plot_qc import qc_metrics

ehrapy/plot/ehrapy_plot/plot_qc.py

@@ -0,0 +1,51 @@
+from __future__ import annotations
+
+from anndata import AnnData
+from rich.console import Console
+from rich.table import Table
+
+
+def qc_metrics(adata: AnnData, extra_columns: list[str] | None = None) -> None:
+    """Plots the calculated quality control metrics for var of adata. Per default this will display the following features:
+    ``missing_values_abs``, ``missing_values_pct``, ``mean``, ``median``, ``standard_deviation``, ``max``, ``min``.
+
+    Args:
+        adata: Annotated data matrix.
+        extra_columns: List of custom (qc) var columns to be displayed additionally.
+
+    """
+    table = Table(title="[bold blue]Ehrapy qc metrics of var")
+    # add special column header for the column name
+    table.add_column("[bold blue]Column name", justify="right", style="bold green")
+    var_names = list(adata.var_names)
+    # default qc columns added to var
+    fixed_qc_columns = [
+        "missing_values_abs",
+        "missing_values_pct",
+        "mean",
+        "median",
+        "standard_deviation",
+        "min",
+        "max",
+    ]
+    # update columns to display with extra columns (if any)
+    columns_to_display = fixed_qc_columns if not extra_columns else fixed_qc_columns + extra_columns
+    # check whether all columns exist (qc has been executed before and extra columns are var columns)
+    if (set(columns_to_display) & set(adata.var.columns)) != set(columns_to_display):
+        raise QCDisplayError(
+            "Cannot display QC metrics of current AnnData object. Either QC has not been executed before or "
+            "some column(s) of the extra_columns parameter are not in var!"
+        )
+    vars_to_display = adata.var[columns_to_display]
+    # add column headers
+    for col in vars_to_display:
+        table.add_column(f"[bold blue]{col}", justify="right", style="bold green")
+    for var in range(len(vars_to_display)):
+        table.add_row(var_names[var], *map(str, list(vars_to_display.iloc[var])))
+
+    console = Console()
+    console.print(table)
+
+
+class QCDisplayError(Exception):
+    pass

ehrapy/preprocessing/__init__.py

@@ -9,6 +9,6 @@
     norm_scale,
     norm_sqrt,
 )
-from ehrapy.preprocessing._quality_control import calculate_qc_metrics
+from ehrapy.preprocessing._quality_control import qc_metrics
 from ehrapy.preprocessing._scanpy_pp_api import *  # noqa: E402,F403
 from ehrapy.preprocessing.encoding._encode import encode, undo_encoding

ehrapy/preprocessing/_data_imputation.py

@@ -398,9 +398,9 @@ def _warn_imputation_threshold(adata: AnnData, var_names: list[str] | None, thre
         adata.var["missing_values_pct"]
     except KeyError:
         print("[bold yellow]Quality control metrics missing. Calculating...")
-        from ehrapy.preprocessing import calculate_qc_metrics
+        from ehrapy.preprocessing import qc_metrics
 
-        calculate_qc_metrics(adata)
+        qc_metrics(adata)
     used_var_names = set(adata.var_names) if var_names is None else set(var_names)
 
     thresholded_var_names = set(adata.var[adata.var["missing_values_pct"] > threshold].index) & set(used_var_names)

ehrapy/preprocessing/_quality_control.py

@@ -7,7 +7,7 @@
 from anndata import AnnData
 
 
-def calculate_qc_metrics(
+def qc_metrics(
     adata: AnnData, qc_vars: Collection[str] = (), layer: str = None, inplace: bool = True
 ) -> pd.DataFrame | None:
     """Calculates various quality control metrics.

tests/preprocessing/test_quality_control.py

@@ -4,7 +4,7 @@
 import pandas as pd
 from anndata import AnnData
 
-from ehrapy.preprocessing._quality_control import _obs_qc_metrics, _var_qc_metrics, calculate_qc_metrics
+from ehrapy.preprocessing._quality_control import _obs_qc_metrics, _var_qc_metrics, qc_metrics
 
 CURRENT_DIR = Path(__file__).parent
 _TEST_PATH = f"{CURRENT_DIR}/test_preprocessing"
@@ -45,7 +45,7 @@ def test_var_qc_metrics(self):
         assert np.allclose(var_metrics["max"].values, np.array([np.nan, np.nan, 41.419998]), equal_nan=True)
 
     def test_calculate_qc_metrics(self):
-        obs_metrics, var_metrics = calculate_qc_metrics(self.test_adata, inplace=True)
+        obs_metrics, var_metrics = qc_metrics(self.test_adata, inplace=True)
 
         assert obs_metrics is not None
         assert var_metrics is not None