Creating CX2Network objects

Below are factories that facilitate creation of CX2Network objects in different ways:

Raw CX2

The RawCX2NetworkFactory is designed to create a CX2Network instance directly from raw CX2 data.

from ndex2.cx2 import RawCX2NetworkFactory, CX2Network

# Sample raw CX2 data
raw_cx2_data = {...}  # Replace with actual raw CX2 data

# Creating an instance of RawCX2NetworkFactory
factory = RawCX2NetworkFactory()

# Creating a CX2Network from raw CX2 data
cx2_network = factory.get_cx2network(raw_cx2_data)

# cx2_network is now a populated CX2Network instance

NetworkX

The NetworkXToCX2NetworkFactory facilitates the conversion of a NetworkX graph into a CX2Network. This allows you to leverage the CX2 format for sharing, storing, or processing networks in Cytoscape Ecosystem.

Features:

Converts nodes, edges, and their attributes from a NetworkX graph to a CX2Network.
Supports layout extraction from node attributes (x, y, z) or external position dictionaries (G.pos and G.zpos).
Handles edge attributes, network-level attributes, and node-level attributes.
Automatically assigns unique IDs to nodes when non-integer IDs are used.

Layout storage settings:

If get_layout_from_pos=True, layout coordinates are extracted from G.pos and G.zpos.
G.pos: A dictionary mapping node IDs to (x, y) tuples
G.zpos: A dictionary mapping node IDs to z-coordinates.
If get_layout_from_pos=False, layout coordinates are extracted from node attributes (x, y, z) if they exist.

Warning

If the Networkx graph contains string node IDs that cannot be converted to integers, the node IDs will be saved in the name attribute. If the graph already has a name attribute for its nodes, this may result in a conflict and loss of information. To avoid this, please rename the existing name attribute before saving the graph.

import networkx as nx
from ndex2.cx2 import NetworkXToCX2NetworkFactory, CX2Network

# Create and populate a NetworkX graph
networkx_graph = nx.MultiDiGraph()
networkx_graph.add_node(1, name='Node1', size=5)
networkx_graph.add_node(2, name='Node2', size=10)
networkx_graph.add_edge(1, 2, weight=1.0, interaction='activates')

# Add layout positions to the graph
networkx_graph.pos = {1: (10.0, 20.0), 2: (15.0, 25.0)}
networkx_graph.zpos = {1: 5.0, 2: 7.0}

# Convert to CX2Network using the factory
factory = NetworkXToCX2NetworkFactory()
cx2_network = factory.get_cx2network(networkx_graph, get_layout_from_pos=True)

# View the CX2Network data
print(cx2_network.to_cx2())

Pandas

The PandasDataFrameToCX2NetworkFactory facilitates the conversion of a pandas.DataFrame into a CX2Network. This is particularly useful for integrating tabular data into the CX2 network structure for visualization, analysis, and sharing in Cytoscape Ecosystem.

Features

Customizable Input: Supports defining custom source and target node fields and node and edge attributes via prefixes or explicit column lists.
Flexible Layout Handling: Automatically detects and uses node layout attributes (x, y, z) when present.
Edge Interactions: Supports custom interaction column name (interaction_col parameter). Default edge interaction type can be set using the edge_interaction parameter.
Default Behavior: If no custom edge attributes are provided, all non-node attribute columns are treated as edge attributes.
Node Attribute Updates: Nodes listed multiple times will have their attributes updated based on the latest occurrence in the DataFrame.

Example 1 (setting node ids)

import pandas as pd
from ndex2.cx2 import PandasDataFrameToCX2NetworkFactory, CX2Network

# DataFrame with source, target, and other columns
data = {'source': [1, 2], 'target': [2, 3],
        'weight': [1.0, 0.9],
        'source_size': [5, 6], 'target_size': [6, 7]}
df = pd.DataFrame(data)

# Creating an instance of PandasDataFrameToCX2NetworkFactory
factory = PandasDataFrameToCX2NetworkFactory()

# Converting DataFrame to CX2Network
cx2_network = factory.get_cx2network(df, source_id='source', target_id='target')

# cx2_network is now a CX2Network instance based on the DataFrame data
print(cx2_network.to_cx2())

Example 2 (using node names)

import pandas as pd
from ndex2.cx2 import PandasDataFrameToCX2NetworkFactory, CX2Network

# DataFrame with source, target, and other columns
data = {'source': ['A', 'B'], 'target': ['B', 'C'],
        'weight': [1.0, 0.9],
        'source_size': [5, 6], 'target_size': [6, 7]}
df = pd.DataFrame(data)

# Creating an instance of PandasDataFrameToCX2NetworkFactory
factory = PandasDataFrameToCX2NetworkFactory()

# Converting DataFrame to CX2Network
cx2_network = factory.get_cx2network(df, source_field='source', target_field='target')

# cx2_network is now a CX2Network instance based on the DataFrame data
print(cx2_network.to_cx2())

Warning

Use source_field and target_field when you use node names (e.g. gene symbols, proteins etc.), use source_id and target_id when you want to set specific node IDs (numerical values). It is possible to use both.

Column Naming Convention

Source and target nodes of an edge.
By default, the columns source_name and target_name represent the names of the source and target nodes, respectively and source_id and target_id represent the unique identifiers for the source and target nodes.

It can be changed by setting the parameter source_field and target_field to column names containing source/ target names, and source_id and target_id to column names containing source/target ids. Specifying ids is not necessary.
Node attributes.
Node attributes can be specified with a prefix according to their node type:
Use the parameter source_node_attr_prefix to set prefix for attributes of the source node (default source_, e.g., source_color, the node will have attribute color).

Use the parameter target_node_attr_prefix to set prefix for attributes of the target node (default target_ e.g., target_size, the node will have attribute size).
If the prefixes are used, attributes in the form prefix_x (e.g. source_x) will be used as x, y, and z coordinates of the node

They can also be explicitly specified as a list passed in parameter source_node_attr for edge source node and target_node_attr for edge target node. The same columns can be used for both source and target node attributes (e.g. source_node_attr=['color', 'size'] and target_node_attr=['color', 'size']).
Edge attributes.
Edge attributes can be specified as list in parameter edge_attr. In case the parameter is not set, all columns that were not used as node attributes will be used (e.g., weight for an edge’s weight attribute).

The edge_interaction parameter defines the default interaction type for edges. If not specified in the data frame as edge attribute, this default value is used. If not set, the default interaction is set to interacts-with.

Example with column names passed as parameters

import pandas as pd
from ndex2.cx2 import PandasDataFrameToCX2NetworkFactory

# DataFrame with custom column names for nodes and attributes
data = {'Protein 1': ['NodeA', 'NodeB'], 'Protein 2': ['NodeB', 'NodeC'],
        'node_id1': [100, 200], 'node_id2': [200, 300], 'connection_strength': [0.8, 0.7],
        'color': ['red', 'red'], 's_size': [1, 2], 't_size': [2, 1]}
df = pd.DataFrame(data)

# Creating an instance of PandasDataFrameToCX2NetworkFactory
factory = PandasDataFrameToCX2NetworkFactory()

# Creating CX2Network with custom parameters
cx2_network_custom = factory.get_cx2network(df,
    source_field='Protein 1', target_field='Protein 2',
    source_id='node_id1', target_id='node_id2',
    source_node_attr=['color', 's_size'], target_node_attr=['color', 't_size'],
    edge_interaction='binds-to')

# cx2_network_custom is now a CX2Network instance with custom settings
print(cx2_network_custom.to_cx2())

Warning

Please note that if a node is listed both as a source and a target, or appears multiple times either as a source or a target, its attributes will be updated to reflect the most recent data entry in the dataframe. This means that each node’s attributes will correspond to their latest occurrence in the dataset.

For example, if node ‘A’ appears in row 1 with the attribute color=red, and then appears again in row 5 of the dataframe with the attribute color=blue, the attribute color of this node will be updated to blue.

NiceCXNetwork

See Convert NiceCXNetwork to CX2Netowrk