scpanel.train

Functions

transform_adata(→ Tuple[anndata._core.anndata.AnnData, ...)

models_train(→ List[Union[Tuple[str, ...)

models_predict(→ Tuple[anndata._core.anndata.AnnData, ...)

models_score(adata_test_final, y_pred_list, ...[, out_dir])

cal_sample_auc(→ numpy.float64)

auc_pvalue(→ float)

pt_pred(→ anndata._core.anndata.AnnData)

pt_score(→ anndata._core.anndata.AnnData)

_panel_grid(→ Tuple[matplotlib.figure.Figure, ...)

plot_roc_curve(→ List[matplotlib.axes._axes.Axes])

convert_pvalue_to_asterisks(→ str)

plot_violin(→ matplotlib.axes._axes.Axes)

Violin Plots for cell-level prediction probabilities in each sample.

make_single_spider(→ None)

Module Contents

scpanel.train.transform_adata(adata_train: anndata._core.anndata.AnnData, adata_test_dict: Dict[str, anndata._core.anndata.AnnData], selected_gene: List[str] | None = None) → Tuple[anndata._core.anndata.AnnData, anndata._core.anndata.AnnData]
scpanel.train.models_train(adata_train_final: anndata._core.anndata.AnnData, search_grid: bool, out_dir: str | None = None, param_grid: Dict[str, Dict[str, int]] | None = None) → List[Tuple[str, sklearn.linear_model._logistic.LogisticRegression] | Tuple[str, sklearn.ensemble._forest.RandomForestClassifier] | Tuple[str, sklearn.svm._classes.SVC] | Tuple[str, sklearn.neighbors._classification.KNeighborsClassifier] | Tuple[str, scpanel.GATclassifier.GATclassifier]]
scpanel.train.models_predict(clfs: List[Tuple[str, sklearn.linear_model._logistic.LogisticRegression] | Tuple[str, sklearn.ensemble._forest.RandomForestClassifier] | Tuple[str, sklearn.svm._classes.SVC] | Tuple[str, sklearn.neighbors._classification.KNeighborsClassifier] | Tuple[str, scpanel.GATclassifier.GATclassifier]], adata_test_final: anndata._core.anndata.AnnData, out_dir: str | None = None) → Tuple[anndata._core.anndata.AnnData, List[Tuple[str, numpy.ndarray] | Tuple[str, torch.Tensor]], List[Tuple[str, numpy.ndarray]]]
scpanel.train.models_score(adata_test_final, y_pred_list, y_pred_score_list, out_dir=None)
scpanel.train.cal_sample_auc(df: pandas.core.frame.DataFrame, score_col: str) → numpy.float64
scpanel.train.auc_pvalue(row: pandas.core.series.Series) → float
scpanel.train.pt_pred(adata_test_final: anndata._core.anndata.AnnData, cell_pred_col: str = 'median_pred_score', num_bootstrap: int | None = None) → anndata._core.anndata.AnnData
scpanel.train.pt_score(adata_test_final: anndata._core.anndata.AnnData, cell_pred_col: str = 'median_pred_score') → anndata._core.anndata.AnnData
scpanel.train._panel_grid(hspace: float, wspace: float, ncols: int, num_panels: int) → Tuple[matplotlib.figure.Figure, matplotlib.gridspec.GridSpec]
scpanel.train.plot_roc_curve(adata_test_final: anndata._core.anndata.AnnData, sample_id: pandas.core.series.Series, cell_pred_col: str, ncols: int = 4, hspace: float = 0.25, wspace: None = None, ax: None = None, scatter_kws: Dict[str, int] | None = None, legend_kws: Dict[str, Dict[str, int]] | None = None) → List[matplotlib.axes._axes.Axes]
Parameters:

  • adata_test_final (-)

  • sample_id (-)

  • cell_pred_col (-)

  • ncols (-)

  • hspace (-)

  • wspace (-)

  • ax (-)

  • scatter_kws (-)

Return type:

Axes

Examples

plot_roc_curve(adata_test_final,

sample_id = [‘C3’,’C6’,’H1’], cell_pred_col = ‘median_pred_score’, scatter_kws={‘s’:10})

scpanel.train.convert_pvalue_to_asterisks(pvalue: float) → str
scpanel.train.plot_violin(adata: anndata._core.anndata.AnnData, cell_pred_col: str = 'median_pred_score', dot_size: int = 2, ax: matplotlib.axes._axes.Axes | None = None, palette: Dict[str, str] | None = None, xticklabels_color: bool = False, text_kws: Dict[Any, Any] = {}) → matplotlib.axes._axes.Axes

Violin Plots for cell-level prediction probabilities in each sample.

Parameters: - adata: AnnData Object

  • cell_pred_col: string, name of the column with cell-level prediction probabilities

in adata.obs (default: ‘median_pred_score’)

  • pt_stat: string, a test for the null hypothesis that the distribution of probabilities

in this sample is different from the population (default: ‘perm’)

Options: - ‘perm’: permutation test - ‘t-test’: one-sample t-test

  • fig_size: tuple, size of figure (default: (10, 3))

  • dot_size: float, Radius of the markers in stripplot.

Returns:

ax

scpanel.train.make_single_spider(adata_test_final: anndata._core.anndata.AnnData, metric_idx: int, color: str, nrow: int, ncol: int) → None