featuremap.features

Created on Wed Nov 1 13:52:14 2023

Functions

create_adata(X, emb_featuremap[, obs, var])

Create an AnnData object for plotting

pseudotime_mst(adata, random_state, start_point_index)

Given a staring point, compute the pseudotime by distance to all other points on the MST

quiver_autoscale(X_emb, V_emb)

Autoscale the arrow plot

plot_gauge(adata[, embedding, vkey, density, smooth, ...])

Plot the gauge embedding to visualize the eigengene (or frame for feature loadings)

plot_gauge_both(adata[, embedding, vkey_1, vkey_2, ...])

Plot the gauge embedding to visualize the eigengene (or frame for feature loadings)

local_intrinsic_dim(adata[, threshold, plot])

Compute the intrinsic dimensionality locally based on local SVD.

feature_variation(adata[, threshold])

Compute the feature variation and feature loadings based on local SVD.

feature_variation_one_feature(adata[, feature, ...])

Compute the feature variation and feature loadings based on local SVD.

feature_projection(adata[, feature, vkey])

Compute the feature variation and feature loadings based on local SVD.

feature_gradient(adata[, feature])

Compute the feature variation and feature loadings based on local SVD.

feature_gradient_projection(adata[, feature, random_state])

plot_feature(adata[, feature, feature_loading_emb, ...])

Plot a given feature (e.g., gene) in two dimensional visualization

plot_one_feature(adata[, feature, embedding, ...])

Plot a given feature (e.g., gene) in two dimensional visualization

plot_multiple_features(adata[, features, embedding, ...])

Plot a given feature (e.g., gene) in two dimensional visualization

plot_one_feature_by_all_cells(adata[, feature, ...])

Plot a given feature (e.g., gene) in two dimensional visualization

plot_multiple_features_by_all_cells(adata[, features, ...])

Plot a given feature (e.g., gene) in two dimensional visualization

plot_multiple_features_by_all_cells_given_clusters(adata)

Plot a given feature (e.g., gene) in two dimensional visualization

variation_feature_pp(adata)

Preprocess the variation feature for DGV analysis

feature_variation_embedding(adata[, n_components, ...])

Compute the feature variation embedding based on all features based on local SVD.

featuremap_var_3d(emb_var_3d[, color, symbol, marker_size])

Plot the feature variation embedding in 3D.

Module Contents

featuremap.features.create_adata(X, emb_featuremap, obs=None, var=None)[source]

Create an AnnData object for plotting

Parameters:

  • X (np.ndarray) – Data matrix.

  • emb_featuremap (object) – Featuremap object.

  • obs (pd.DataFrame)

  • var (pd.DataFrame)

Returns:

adata – Annotated data matrix.

Return type:

AnnData

featuremap.features.pseudotime_mst(adata, random_state, start_point_index)[source]

Given a staring point, compute the pseudotime by distance to all other points on the MST

Parameters:

  • adata (AnnData) – Annotated data matrix.

  • random_state (int) – Random seed.

  • start_point_index (int) – Starting point index.

Returns:

pseudotime_mst : np.ndarray. Pseudotime based on MST.

Return type:

adata.obs[‘feat_pseudotime’] = pseudotime_mst

featuremap.features.quiver_autoscale(X_emb, V_emb)[source]

Autoscale the arrow plot

Parameters:

  • X_emb (np.ndarray) – Embedding matrix.

  • V_emb (np.ndarray) – Variation embedding matrix.

Returns:

Scale factor for quiver plot.

Return type:

Q.scale / scale_factor * 5

featuremap.features.plot_gauge(adata: anndata.AnnData, embedding='X_featmap', vkey='gauge_v1_emb', density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True)[source]

Plot the gauge embedding to visualize the eigengene (or frame for feature loadings)

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • vkey (string) – Variation key. The default is ‘gauge_v1_emb’.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

featuremap.features.plot_gauge_both(adata: anndata.AnnData, embedding='X_featmap', vkey_1='gauge_v1_emb', vkey_2='gauge_v2_emb', density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True, scale_quiver=1)[source]

Plot the gauge embedding to visualize the eigengene (or frame for feature loadings)

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

featuremap.features.local_intrinsic_dim(adata: anndata.AnnData, threshold=0.9, plot: bool = False)[source]

Compute the intrinsic dimensionality locally based on local SVD.

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • threshold (float) – Threshold for proportion of variance contributions. The default is 0.9.

  • plot (bool) – If True, plot a histogram of the intrinsic dimensionality. Default False.

Returns:

intrinsic_dim – Intrinsic dimensionality.

Return type:

np.ndarray

featuremap.features.feature_variation(adata: anndata.AnnData, threshold=0.5)[source]

Compute the feature variation and feature loadings based on local SVD.

Parameters:

adata (AnnData) – An annotated data matrix.

featuremap.features.feature_variation_one_feature(adata: anndata.AnnData, feature='', threshold=0.9, feature_key='variation_feature')[source]

Compute the feature variation and feature loadings based on local SVD.

Parameters:

adata (AnnData) – An annotated data matrix.

featuremap.features.feature_projection(adata: anndata.AnnData, feature='', vkey='VH_embedding')[source]

Compute the feature variation and feature loadings based on local SVD.

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

featuremap.features.feature_gradient(adata: anndata.AnnData, feature='')[source]

Compute the feature variation and feature loadings based on local SVD.

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

featuremap.features.feature_gradient_projection(adata: anndata.AnnData, feature='', random_state=42)[source]
featuremap.features.plot_feature(adata: anndata.AnnData, feature='', feature_loading_emb='feature_loading_embedding', embedding='X_featmap', cluster_key='clusters', plot_within_cluster=[], pseudotime_adjusted=False, pseudotime='dpt_pseudotime', trend='positive', ratio=0.2, density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True)[source]

Plot a given feature (e.g., gene) in two dimensional visualization

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • cluster_key (string) – Cluster name indicator. The default is ‘clusters’.

  • plot_within_cluster (list) – A list of clusters in which the feaure is to plot. The default is [].

  • pseudotime_adjusted (bool) – Whether to adjust the feature direction by pseudotime. The default is False.

  • pseudotime (string) – Pseudotime indicator. The default is ‘dpt_pseudotime’.

  • trend (string of {'positive','negative'}) – The direction along pseudotime. The default is ‘positive’.

  • ratio (float) – Filtering ratio by expression to filter varition by low expression. The default is 0.5.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

featuremap.features.plot_one_feature(adata: anndata.AnnData, feature='', embedding='X_featmap', cluster_key='clusters', plot_within_cluster=[], pseudotime_adjusted=False, pseudotime='dpt_pseudotime', trend='positive', ratio=0.2, density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True, scale_quiver=1.0, output_v_grid=False, scale_by_variation=False, ax=None)[source]

Plot a given feature (e.g., gene) in two dimensional visualization

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • cluster_key (string) – Cluster name indicator. The default is ‘clusters’.

  • plot_within_cluster (list) – A list of clusters in which the feaure is to plot. The default is [].

  • pseudotime_adjusted (bool) – Whether to adjust the feature direction by pseudotime. The default is False.

  • pseudotime (string) – Pseudotime indicator. The default is ‘dpt_pseudotime’.

  • trend (string of {'positive','negative'}) – The direction along pseudotime. The default is ‘positive’.

  • ratio (float) – Filtering ratio by expression to filter varition by low expression. The default is 0.5.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

  • output_v_grid (bool) – Output the grid velocity. The default is False.

  • scale_by_variation (bool) – Scale the arrow plot by feature variation. The default is True.

  • ax (matplotlib axis) – The axis to plot the feature. The default is None.

featuremap.features.plot_multiple_features(adata: anndata.AnnData, features=[], embedding='X_featmap', cluster_key='clusters', plot_within_cluster=[], pseudotime_adjusted=False, pseudotime='dpt_pseudotime', trend='positive', ratio=0.2, density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True, scale_by_variation=True)[source]

Plot a given feature (e.g., gene) in two dimensional visualization

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • cluster_key (string) – Cluster name indicator. The default is ‘clusters’.

  • plot_within_cluster (list) – A list of clusters in which the feaure is to plot. The default is [].

  • pseudotime_adjusted (bool) – Whether to adjust the feature direction by pseudotime. The default is False.

  • pseudotime (string) – Pseudotime indicator. The default is ‘dpt_pseudotime’.

  • trend (string of {'positive','negative'}) – The direction along pseudotime. The default is ‘positive’.

  • ratio (float) – Filtering ratio by expression to filter varition by low expression. The default is 0.5.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

  • output_v_grid (bool) – Output the grid velocity. The default is False.

  • scale_by_variation (bool) – Scale the arrow plot by feature variation. The default is True.

featuremap.features.plot_one_feature_by_all_cells(adata: anndata.AnnData, feature='', embedding='X_featmap', cluster_key='clusters', plot_within_cluster=[], pseudotime_adjusted=False, pseudotime='dpt_pseudotime', trend='positive', ratio=0.2, density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True, output_v_grid=False, scale_by_variation=True, scale_quiver=1.0)[source]

Plot a given feature (e.g., gene) in two dimensional visualization

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • cluster_key (string) – Cluster name indicator. The default is ‘clusters’.

  • plot_within_cluster (list) – A list of clusters in which the feaure is to plot. The default is [].

  • pseudotime_adjusted (bool) – Whether to adjust the feature direction by pseudotime. The default is False.

  • pseudotime (string) – Pseudotime indicator. The default is ‘dpt_pseudotime’.

  • trend (string of {'positive','negative'}) – The direction along pseudotime. The default is ‘positive’.

  • ratio (float) – Filtering ratio by expression to filter varition by low expression. The default is 0.5.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

featuremap.features.plot_multiple_features_by_all_cells(adata: anndata.AnnData, features=[], embedding='X_featmap', cluster_key='clusters', plot_within_cluster=[], pseudotime_adjusted=False, pseudotime='dpt_pseudotime', trend='positive', ratio=0.2, density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True, output_v_grid=False, scale_by_variation=True)[source]

Plot a given feature (e.g., gene) in two dimensional visualization

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • cluster_key (string) – Cluster name indicator. The default is ‘clusters’.

  • plot_within_cluster (list) – A list of clusters in which the feaure is to plot. The default is [].

  • pseudotime_adjusted (bool) – Whether to adjust the feature direction by pseudotime. The default is False.

  • pseudotime (string) – Pseudotime indicator. The default is ‘dpt_pseudotime’.

  • trend (string of {'positive','negative'}) – The direction along pseudotime. The default is ‘positive’.

  • ratio (float) – Filtering ratio by expression to filter varition by low expression. The default is 0.5.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

featuremap.features.plot_multiple_features_by_all_cells_given_clusters(adata: anndata.AnnData, features=[], embedding='X_featmap', cluster_key='clusters', plot_within_cluster=[], pseudotime_adjusted=True, pseudotime='feat_pseudotime', trend='positive', ratio=0.2, density=1, smooth=0.5, n_neighbors=None, min_mass=1, autoscale=True, scale_by_variation=True, feature_key='', scale_quiver=1.0)[source]

Plot a given feature (e.g., gene) in two dimensional visualization

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • feature (string) – Feature name to be plotted.

  • embedding (string) – Embedding background for feature plot. The default is ‘X_featmap’.

  • cluster_key (string) – Cluster name indicator. The default is ‘clusters’.

  • plot_within_cluster (list) – A list of clusters in which the feaure is to plot. The default is [].

  • pseudotime_adjusted (bool) – Whether to adjust the feature direction by pseudotime. The default is False.

  • pseudotime (string) – Pseudotime indicator. The default is ‘dpt_pseudotime’.

  • trend (string of {'positive','negative'}) – The direction along pseudotime. The default is ‘positive’.

  • ratio (float) – Filtering ratio by expression to filter varition by low expression. The default is 0.5.

  • density (float) – Grid desity for plot. The default is 1.

  • smooth (float) – For kde estimation. The default is 0.5.

  • n_neighbors (int) – Number of neighbours for kde. The default is None.

  • min_mass (float) – Minumum denstiy to show the grid plot. The default is 1.

  • autoscale (bool) – Scale the arrow plot. The default is True.

featuremap.features.variation_feature_pp(adata)[source]

Preprocess the variation feature for DGV analysis

Parameters:

adata (AnnData) – An annotated data matrix.

featuremap.features.feature_variation_embedding(adata, n_components=2, layer='variation_feature', variation_preprocess_flag=False, random_state=42)[source]

Compute the feature variation embedding based on all features based on local SVD.

Parameters:

  • adata (AnnData) – An annotated data matrix.

  • n_components (int) – Number of components for embedding. The default is 2.

  • layer (string) – Layer for variation feature. The default is ‘variation_feature’.

  • variation_preprocess_flag (bool) – Whether to preprocess the variation feature. The default is False.

  • random_state (int) – Random state. The default is 42.

Returns:

adata_var – Annotated data matrix with variation matrix and variation embedding.

Return type:

AnnData

featuremap.features.featuremap_var_3d(emb_var_3d, color=None, symbol=None, marker_size=3)[source]

Plot the feature variation embedding in 3D.

Parameters:

  • emb_var_3d (np.ndarray) – 3D embedding of feature variation.

  • color (string) – Color indicator. The default is None.

  • symbol (string) – Symbol indicator. The default is None.

  • marker_size (int) – Marker size. The default is 3.