Attributes, Views, and Indexing¶

This notebook covers the parts of the API that make AnnNet pleasant to inspect:

• attribute upserts for vertices, edges, slices, and graph-level metadata
• read-only dataframe-style views
• index helpers for mapping IDs to incidence rows and columns
• GraphView for filtered subgraph exploration

In [1]:

from pathlib import Path
import sys

NOTEBOOK_DIR = Path.cwd()
PACKAGE_ROOT = (NOTEBOOK_DIR / "..").resolve()
if str(PACKAGE_ROOT) not in sys.path:
    sys.path.insert(0, str(PACKAGE_ROOT))

import annnet as an

from pathlib import Path import sys NOTEBOOK_DIR = Path.cwd() PACKAGE_ROOT = (NOTEBOOK_DIR / "..").resolve() if str(PACKAGE_ROOT) not in sys.path: sys.path.insert(0, str(PACKAGE_ROOT)) import annnet as an

In [2]:

from annnet.core._Views import GraphView

G = an.AnnNet(directed=True)
G.add_vertices(["EGFR", "GRB2", "SOS1", "RAS"])
G.add_edge("EGFR", "GRB2", edge_id="e1")
G.add_edge("GRB2", "SOS1", edge_id="e2")
G.add_edge("SOS1", "RAS", edge_id="e3")

G.set_graph_attribute("dataset", "toy receptor signaling")
G.set_vertex_attrs("EGFR", family="RTK", compartment="membrane")
G.set_vertex_attrs("GRB2", family="adapter", compartment="cytosol")
G.set_edge_attrs("e1", relation="recruits", evidence="literature")
G.set_edge_attrs("e2", relation="binds")
G.set_edge_attrs("e3", relation="activates")

from annnet.core._Views import GraphView G = an.AnnNet(directed=True) G.add_vertices(["EGFR", "GRB2", "SOS1", "RAS"]) G.add_edge("EGFR", "GRB2", edge_id="e1") G.add_edge("GRB2", "SOS1", edge_id="e2") G.add_edge("SOS1", "RAS", edge_id="e3") G.set_graph_attribute("dataset", "toy receptor signaling") G.set_vertex_attrs("EGFR", family="RTK", compartment="membrane") G.set_vertex_attrs("GRB2", family="adapter", compartment="cytosol") G.set_edge_attrs("e1", relation="recruits", evidence="literature") G.set_edge_attrs("e2", relation="binds") G.set_edge_attrs("e3", relation="activates")

In [3]:

print("graph attributes:", G.get_graph_attributes())
print("EGFR attrs:", G.get_vertex_attrs("EGFR"))
print("e1 attrs:", G.get_edge_attrs("e1"))

print("graph attributes:", G.get_graph_attributes()) print("EGFR attrs:", G.get_vertex_attrs("EGFR")) print("e1 attrs:", G.get_edge_attrs("e1"))

graph attributes: {'dataset': 'toy receptor signaling'}
EGFR attrs: {'vertex_id': 'EGFR', 'family': 'RTK', 'compartment': 'membrane'}
e1 attrs: {'edge_id': 'e1', 'relation': 'recruits', 'evidence': 'literature'}

Read-only dataframe views¶

These helpers expose the current attribute state without changing the graph.

In [4]:

G.vertices_view()

G.vertices_view()

Out[4]:

shape: (4, 3)

vertex_id	family	compartment
str	str	str
"EGFR"	"RTK"	"membrane"
"GRB2"	"adapter"	"cytosol"
"SOS1"	null	null
"RAS"	null	null

In [5]:

G.edges_view()

G.edges_view()

Out[5]:

shape: (3, 13)

edge_id	kind	directed	global_weight	source	target	edge_type	head	tail	members	relation	evidence	effective_weight
str	str	bool	f64	str	str	str	list[str]	list[str]	list[str]	str	str	f64
"e1"	"binary"	true	1.0	"EGFR"	"GRB2"	"regular"	null	null	null	"recruits"	"literature"	1.0
"e2"	"binary"	true	1.0	"GRB2"	"SOS1"	"regular"	null	null	null	"binds"	null	1.0
"e3"	"binary"	true	1.0	"SOS1"	"RAS"	"regular"	null	null	null	"activates"	null	1.0

Index helpers¶

G.idx is the public namespace for row and column lookups in the incidence matrix.

In [6]:

print("EGFR row:", G.idx.entity_to_row("EGFR"))
print("row 0 ->", G.idx.row_to_entity(0))
print("e2 column:", G.idx.edge_to_col("e2"))
print("index stats:", G.idx.stats())

print("EGFR row:", G.idx.entity_to_row("EGFR")) print("row 0 ->", G.idx.row_to_entity(0)) print("e2 column:", G.idx.edge_to_col("e2")) print("index stats:", G.idx.stats())

EGFR row: 0
row 0 -> EGFR
e2 column: 1
index stats: {'n_entities': 4, 'n_vertices': 4, 'n_edge_entities': 0, 'n_edges': 3, 'max_row': 3, 'max_col': 2}

Filtered graph views¶

GraphView is the lazy filtered view object. It does not duplicate the graph unless you explicitly materialize it.

In [7]:

signal_view = GraphView(G, vertices=["EGFR", "GRB2", "SOS1"])

print("vertex ids:", sorted(signal_view.vertex_ids))
print("edge ids:", signal_view.edge_ids)
signal_view.obs

signal_view = GraphView(G, vertices=["EGFR", "GRB2", "SOS1"]) print("vertex ids:", sorted(signal_view.vertex_ids)) print("edge ids:", signal_view.edge_ids) signal_view.obs

vertex ids: ['EGFR', 'GRB2', 'SOS1']
edge ids: None

Out[7]:

shape: (3, 3)

vertex_id	family	compartment
str	str	str
"EGFR"	"RTK"	"membrane"
"GRB2"	"adapter"	"cytosol"
"SOS1"	null	null

In [8]:

materialized = signal_view.materialize()
print(materialized.shape)
materialized.edges_view()

materialized = signal_view.materialize() print(materialized.shape) materialized.edges_view()

(3, 2)

Out[8]:

shape: (2, 13)

edge_id	kind	directed	global_weight	source	target	edge_type	head	tail	members	relation	evidence	effective_weight
str	str	bool	f64	str	str	str	list[str]	list[str]	list[str]	str	str	f64
"edge_0"	"binary"	true	1.0	"EGFR"	"GRB2"	"regular"	null	null	null	"recruits"	"literature"	1.0
"edge_1"	"binary"	true	1.0	"GRB2"	"SOS1"	"regular"	null	null	null	"binds"	null	1.0

In [ ]: