tfcomb.network module

tfcomb.network.build_network(edge_table, node1='TF1', node2='TF2', node_table=None, directed=False, multi=False, tool='networkx', verbosity=1)

Build a network object from a table using either ‘networkx’ or ‘graph-tool’.

Parameters:

• edge_table (pd.DataFrame) – Table containing rows of edges and edge information between node1/node2.

• node1 (str, optional) – The column to use as node1 ID. Default: “TF1”.

• node2 (str, optional) – The column to use as node2 ID. Default: “TF2”.

• node_table (pandas.DataFrame) – A table of attributes to use for nodes. Default: node attributes are estimated from the columns in edge_table.

• directed (bool, optional) – Whether edges are directed or not. Default: False.

• multi (bool, optional) – Allow multiple edges between two vertices. NOTE: Only valid for tool == ‘networkx’. If False, the first occurrence of TF1-TF2/TF2-TF1 in the table is used. Default: False.

• tool (str, optional) – Which module to use for generating network. Must be one of ‘networkx’ or ‘graph-tool’. Default: ‘networkx’.

• verbosity (int, optional) – Verbosity of logging (0/1/2/3). Default: 1.

Returns:

• if tool is ‘networkx’ (networkx.Graph / networkx.DiGraph / networkx.MultiGraph / networkx.MultiDiGraph - depending on parameters given.)

• if tool is ‘graph-tool’ (graph_tool.Graph)

tfcomb.network.get_degree(G, weight=None, direction='both')

Get degree per node in graph. If weight is given, the degree is the sum of weighted edges.

Parameters:

• G (networkx.Graph) – An instance of networkx.Graph

• weight (str, optional) – Name of an edge attribute within network. Default: None.

• direction (str, optional) – Which edge direction to use for calculating degrees. Can be one of: [“both”, “in”, “out”]. Default: ‘both’.

Returns:

A table of format (…)

Return type:

DataFrame

tfcomb.network.get_betweenness_centrality(G, weight=None)

Parameters:

• G (networkx.Graph) – An instance of networkx.Graph

• weight – Edge attribute. Default: None.

tfcomb.network.subset_graph(G, nodes, depth=0)

Subset a graph to a subset of nodes and their neighborhoods at the depth given by ‘depth’.

Parameters:

• G (networkx.Graph) – An instance of networkx.Graph.

• nodes (str or list of str) – Nodes to keep.

• depth (int, optional) – Default: 0 (only edges between given nodes)

tfcomb.network.cluster_louvain(G, weight=None, attribute_name='cluster', logger=None)

Cluster a network using community louvain clustering. By default, sets the attribute “cluster” to each node.

Parameters:

• G (networkx.Graph) – An instance of a network graph to cluster.

• weight (str) – Attribute in graph to use as weight. The higher the weight, the stronger the link. Default: None.

• attribute_name (str) – The attribute name to use for saving clustering. Default: “cluster”.

• logger (a logger object) – An instance of a logger. Default: No logging.

Return type:

None - clustering is added to ‘G’ in place.

tfcomb.network.cluster_blockmodel(g, attribute_name='cluster')

Clustering of a graph-tool graph using stochastic block model minimization.

Parameters:

• g (a graph.tool graph) – An instance of a graph.tool graph.

• attribute_name (str) – The attribute name to use for saving clustering. Default: “cluster”.

tfcomb.network.get_node_table(G)

Get a table containing node names and node attributes for G.

Parameters:

G (a networkx Graph object or graph_tool Graph object) –

Return type:

pandas.DataFrame

tfcomb.network.get_edge_table(G)

Get a table containing edge names and edge attributes for G.

Parameters:

G (a networkx Graph object.) –

Return type:

pandas.DataFrame

tfcomb.network.create_random_network(nodes, edges)

Create a random network with the given list of nodes and total number of edges.

Parameters:

• nodes (list) – List of nodes to use in network.

• edges (int) – Number of edges between nodes.

Return type:

networkx.Graph containing random edges between nodes.

tfcomb.network.plot_powerlaw(G, title='Node degree powerlaw fit', color='blue', save=None)

Fit and plot a powerlaw distribution to the node degrees in the network.

Parameters:

• G (a networkx Graph object) – Networkx containing nodes and edges to analyze.

• title (str, optional) – The title of the resulting plot. Default: “Node degree powerlaw fit”.

• color (str, optional) – The color of the data plotted. Default: “blue”.

• save (str, optional) – If not None, save the plot to the given path. Default: None.

Returns:

ax – Axes object containing the plot.

Return type:

matplotlib.axes