Loading a Network¶

RISK includes dedicated loader functions for multiple network formats. Each returns a standardized networkx graph object for downstream analysis.

Supported Input Formats¶

Format	Function
`.cys`	`load_network_cytoscape()`
`.cyjs`	`load_network_cyjs()`
`.gpickle`	`load_network_gpickle()`
`NetworkX`	`load_network_networkx()`

Each loader supports optional spherical projection, depth tuning, and node filtering.

Cytoscape `.cys` Files¶

Use this format if you've exported or styled your network in Cytoscape.

network = risk.load_network_cytoscape(
    filepath="./data/cytoscape/michaelis_2023.cys",
    source_label="source",
    target_label="target",
    view_name="",
    compute_sphere=True,
    surface_depth=0.1,
)

source_label, target_label: Column names for edges
view_name: Load a specific layout (optional)
compute_sphere: Project layout onto a 3D sphere
surface_depth: Controls visual node "depth"

Cytoscape JSON (`.cyjs`) Files¶

Structured network files exported from Cytoscape Web or JS-based tools.

network = risk.load_network_cyjs(
    filepath="./data/cyjs/michaelis_2023.cyjs",
    source_label="source",
    target_label="target",
    compute_sphere=True,
    surface_depth=0.1,
    min_edges_per_node=0,
)

min_edges_per_node: Filters out sparsely connected nodes

GPickle (`.gpickle`) Files¶

Fast, native Python serialization of NetworkX graphs.

network = risk.load_network_gpickle(
    filepath="./data/gpickle/michaelis_2023.gpickle",
    compute_sphere=True,
    surface_depth=0.1,
    min_edges_per_node=0,
)

Use this for reproducibility and performance when working with saved graphs.

NetworkX Graphs¶

Load an in-memory NetworkX graph directly.

network = risk.load_network_networkx(
    network=network,
    compute_sphere=True,
    surface_depth=0.1,
    min_edges_per_node=0,
)

Useful if you've already constructed a graph using other tools or workflows.

Spherical Projection and Depth¶

All formats support these shared preprocessing parameters:

compute_sphere=True: Projects nodes from 2D to a 3D spherical layout (Mercator-inspired)
surface_depth: Push or pull nodes inward/outward to reflect density or importance
min_edges_per_node: Removes low-degree noise

These features improve layout clarity and biological interpretability.

Next Step¶

Proceed to 3. Annotation to map biological terms onto network nodes.