Backend accessors¶

AnnNet can remain the source of record while dispatching to installed graph, dataframe, and plotting backends.

In [1]:

import annnet as an

import annnet as an

Inspect installed optional components¶

In [2]:

print('graph backends:', an.available_backends())
print('dataframe backends:', an.available_dataframe_backends())
print('plot backends:', an.available_plot_backends())
print('selected dataframe backend:', an.select_dataframe_backend('auto'))

print('graph backends:', an.available_backends()) print('dataframe backends:', an.available_dataframe_backends()) print('plot backends:', an.available_plot_backends()) print('selected dataframe backend:', an.select_dataframe_backend('auto'))

graph backends: {'networkx': True, 'igraph': True, 'graph-tool': False, 'pyg': True}
dataframe backends: {'polars': True, 'pandas': True, 'pyarrow': True}
plot backends: {'graphviz': True, 'pydot': False, 'matplotlib': True}
selected dataframe backend: polars

NetworkX and igraph access¶

In [3]:

G = an.AnnNet(directed=True)
G.add_vertices(['A', 'B', 'C', 'D'])
G.add_edges('A', 'B', edge_id='e1', weight=1.0)
G.add_edges('A', 'B', edge_id='e1_parallel', weight=0.6)
G.add_edges('B', 'C', edge_id='e2', weight=1.4)
G.add_edges('C', 'D', edge_id='complex_cd', directed=False, weight=0.8)

nx_graph = G.nx.backend()
ig_graph, manifest = an.to_igraph(G)
restored = an.from_igraph(ig_graph, manifest)

print(type(nx_graph).__name__, sorted(nx_graph.nodes()))
print('NetworkX edges:', list(nx_graph.edges(data=True)))
print('igraph:', ig_graph.vcount(), 'vertices,', ig_graph.ecount(), 'edges')
print('igraph edge list:', ig_graph.get_edgelist())
print('round-trip:', restored.shape)

G = an.AnnNet(directed=True) G.add_vertices(['A', 'B', 'C', 'D']) G.add_edges('A', 'B', edge_id='e1', weight=1.0) G.add_edges('A', 'B', edge_id='e1_parallel', weight=0.6) G.add_edges('B', 'C', edge_id='e2', weight=1.4) G.add_edges('C', 'D', edge_id='complex_cd', directed=False, weight=0.8) nx_graph = G.nx.backend() ig_graph, manifest = an.to_igraph(G) restored = an.from_igraph(ig_graph, manifest) print(type(nx_graph).__name__, sorted(nx_graph.nodes())) print('NetworkX edges:', list(nx_graph.edges(data=True))) print('igraph:', ig_graph.vcount(), 'vertices,', ig_graph.ecount(), 'edges') print('igraph edge list:', ig_graph.get_edgelist()) print('round-trip:', restored.shape)

MultiDiGraph ['A', 'B', 'C', 'D']
NetworkX edges: [('A', 'B', {'weight': 1.0}), ('A', 'B', {'weight': 0.6}), ('B', 'C', {'weight': 1.4}), ('C', 'D', {'weight': 0.8}), ('D', 'C', {'weight': 0.8})]
igraph: 4 vertices, 5 edges
igraph edge list: [(3, 1), (3, 1), (1, 0), (0, 2), (2, 0)]
round-trip: (4, 4)

Draw the AnnNet source graph¶

In [4]:

from annnet.utils import plotting

plotting.plot(G, backend='graphviz', show_edge_labels=True)

from annnet.utils import plotting plotting.plot(G, backend='graphviz', show_edge_labels=True)

Out[4]:

Optional backend accessors¶

The same accessor pattern applies to other installed graph backends. AnnNet keeps the annotated graph state and projects it when another library has the algorithm or file format you need.

Backend accessors are execution targets, not replacement graph stores. AnnNet keeps the annotated graph state and projects it when another library has the algorithm or file format you need.