ddn3.visualize2

Visualization functions of DDN, improved v2 version

These functions are still quite basic. For advanced plotting of networks, consider using specialized tools.

Module Contents

Functions

draw_network_for_ddn(edges_df, nodes_to_draw[, mode, ...])

Draw networks for DDN

create_nx_graph(edges_df, nodes_show[, min_alpha, ...])

Create NetworkX graph based on edge data frame.

plot_network(G, pos, d_min, labels, node_type, ...[, ...])

Draw the network

get_pos_multi_parts(nodes_all, nodes_type[, cen_lst, ...])

Find positions of nodes with multiple isolated subgraphs.

_get_pos_one_part(nodes_show[, cen, rad, pos])

Find positions of nodes on a circle

_draw_network_labels(ax, pos, d_min, node_type, ...)

Add labels to a graph

_add_node_to_a_circle(pos, nodes, cen, rad, angles)

Find positions in an ellipse, and calculate the minimum distances between points

_angles_in_ellipse(num, a, b)

Calculate angles of evenly spaced points in an ellipse
ddn3.visualize2.draw_network_for_ddn(edges_df, nodes_to_draw, mode='common', nodes_type=None, cen_lst=None, rad_lst=None, labels=None, fig_size=(16, 9), font_size_scale=1, node_size_scale=1, pdf_name='')

Draw networks for DDN

Support drawing any number of ellipses according to nodes_type. The positions and shapes are given in cen_lst and rad_lst. This makes the layout of the graphs more flexible.

The direction of the labels now points to the center of each ellipse. This allows showing larger fonts.

The node size, font size, edge weight are now automatically adjusted according to figure size and node number.

For common network, if two nodes in an edge have same type, draw grey line. If two nodes in an edge have different type, draw green line.

For differential network, if an edge comes from condition 1, draw blue line. If an edge comes from condition 2, draw red line.

Parameters:

  • edges_df (pandas.DataFrame) – Edge information. First two columns for the two feature names. Third column for edge type (common=0, diff1=1, diff2=2) Fourth column for weight.

  • nodes_to_draw (list of str) – Name of nodes to draw

  • fig_size (tuple, optional) – Size of figure.

  • font_size_scale (int, optional) – Scale of fonts, by default 1 To make the font larger, set larger value. The default value is scaled according to the node number.

  • node_size_scale (int, optional) – Scale of node sizes, by default 2

  • nodes_type (None or dict, optional) – Node type (e.g., Gene=0, TF=1), by default None

  • cen_lst (None or ndarray, optional) – The center of ellipse for each type of node. For node type i, use cen_lst[i] Shape (k, 2), k is the number of types. 2 for (x, y) Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • rad_lst (None or ndarray, optional) – The radius of ellipse for each type of node. For node type i, use rad_lst[i] Shape (k, 2), k is the number of types. 2 for shorter and longer axis length. Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • labels (dict) – Alternative (e.g., simplified) names for nodes

  • mode (str, optional) – Draw common graph or differential graph, by default “common”

  • pdf_name (str, optional) – Name of the PDF file to export, by default “”. If set as “”, no pdf will be output.

Returns:

Graph object

Return type:

nx.Graph

ddn3.visualize2.create_nx_graph(edges_df, nodes_show, min_alpha=0.2, max_alpha=1.0, mode='common', nodes_type=None)

Create NetworkX graph based on edge data frame.

Add nodes and edges. Provide visualization related properties to the nodes.

For common network, if two nodes in an edge have same type, draw grey line. If two nodes in an edge have different type, draw green line.

For differential network, if an edge comes from condition 1, draw blue line. If an edge comes from condition 2, draw red line.

Parameters:

  • edges_df (pandas.DataFrame) – Edge information. First two columns for the two feature names. Third column for edge type (common=0, diff1=1, diff2=2) Fourth column for weight.

  • nodes_show (list of str) – Name of nodes to draw

  • min_alpha (float, optional) – Minimum alpha value of edges, by default 0.2 This is for the most light edges.

  • max_alpha (float, optional) – Maximum alpha value of edges, by default 1.0

  • mode (str, optional) – Draw common graph or differential graph, by default “common”

  • nodes_type (None or dict, optional) – Node type (e.g., Gene=0, TF=1), by default None

Returns:

Generated graph

Return type:

nx.Graph

ddn3.visualize2.plot_network(G, pos, d_min, labels, node_type, cen_lst, fig_size, font_size_scale=1, node_size_scale=2, font_alpha_min=0.4)

Draw the network

Parameters:

  • G (nx.Graph) – Graph to draw

  • pos (dict) – Position of each node

  • d_min (float) – Minimum distance between nodes. We use this to adjust node size, font size, etc.

  • labels (dict) – Alternative names for nodes

  • node_type (None or dict, optional) – Node type (e.g., Gene=0, TF=1), by default None

  • cen_lst (None or ndarray, optional) – The center of ellipse for each type of node. For node type i, use cen_lst[i] Shape (k, 2), k is the number of types. 2 for (x, y) Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • fig_size (tuple) – Size of figure. The unit is inch.

  • font_size_scale (int, optional) – Scale of fonts, by default 1 To make the font larger, set larger value. The default value is scaled according to the node number.

  • node_size_scale (int, optional) – Scale of node sizes, by default 1

  • font_alpha_min (float, optional) – The smallest alpha value for fonts in labels, between 0 and 1

ddn3.visualize2.get_pos_multi_parts(nodes_all, nodes_type, cen_lst=np.array([[-0.6, 0], [0.6, 0]]), rad_lst=np.array([[0.4, 1], [0.4, 1]]))

Find positions of nodes with multiple isolated subgraphs.

Each subgraph is an ellipse. Also provides the minimum distances between nodes.

Parameters:

  • nodes_all (list of str) – Node names

  • nodes_type (dict) – feature type ID for each node

  • cen_lst (None or ndarray, optional) – The center of ellipse for each type of node. For node type i, use cen_lst[i] Shape (k, 2), k is the number of types. 2 for (x, y) Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • rad_lst (None or ndarray, optional) – The radius of ellipse for each type of node. For node type i, use rad_lst[i] Shape (k, 2), k is the number of types. 2 for shorter and longer axis length. Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

Returns:

  • pos (dict) – The positions for each node

  • d_min (float) – Minimum distances between nodes

ddn3.visualize2._get_pos_one_part(nodes_show, cen=(0.0, 0.0), rad=(1.0, 1.0), pos=None)

Find positions of nodes on a circle

Also provides the minimum distances between nodes.

Parameters:

  • nodes_show (tuple of str) – Node names

  • cen (ndarray, optional) – The center of ellipse for each type of node. Shape (2, ) Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • rad (ndarray, optional) – The radius of ellipse for each type of node. Shape (2, ) Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • pos (dict) – NetworkX position dictionary

Returns:

  • pos (dict) – The positions for each node

  • d_min (float) – Minimum distances between nodes

ddn3.visualize2._draw_network_labels(ax, pos, d_min, node_type, cen_lst, labels, font_size_lst, font_alpha_lst)

Add labels to a graph

Modified from Networkx add label function

Parameters:

  • ax (matplotlib.axes.Axes) – The axes of the figure

  • pos (dict) – NetworkX position dictionary

  • d_min (float) – Minimum distance between nodes

  • node_type (dict) – Feature type index of each node

  • cen_lst (None or ndarray, optional) – The center of ellipse for each type of node. For node type i, use cen_lst[i] Shape (k, 2), k is the number of types. 2 for (x, y) Assume the y-axis of the figure is in range [-1,1], so do not set too large values.

  • labels (dict) – Alternative names for each node, to be shown in the graph

  • font_size_lst (array_like) – Font size for each node, in points

  • font_alpha_lst (array_like) – Font alpha for each node, value from 0 to 1

ddn3.visualize2._add_node_to_a_circle(pos, nodes, cen, rad, angles)

Find positions in an ellipse, and calculate the minimum distances between points

Parameters:

  • pos (dict) – For saving the position results for NetworkX

  • nodes (list of str) – Name of nodes

  • cen (array_like) – Central position of this circle, shape (2,)

  • rad (array_like) – Length of two axes of the ellipse, shape (2,)

  • angles (array_like) – Angles of the points

Returns:

Minimum distances between points

Return type:

float

ddn3.visualize2._angles_in_ellipse(num, a, b)

Calculate angles of evenly spaced points in an ellipse

Based on https://stackoverflow.com/a/52062369, which is from https://pypi.org/project/flyingcircus/

Parameters:

  • num (int) – Sample number to get

  • a (float) – Length of shorter axis

  • b (float) – Length of longer axis

Returns:

angles – Angles of sampled points

Return type:

ndarray