gpgraph.base¶

The core graph class, plus the I/O class-method constructors.

GenotypePhenotypeGraph¶

class GenotypePhenotypeGraph(nx.DiGraph):
    def __init__(self) -> None: ...

A NetworkX directed graph wrapping a gpmap-v2 GenotypePhenotypeMap. Construction is only supported through from_gpm; the v1 add_gpm post-construction injection is removed.

from_gpm¶

@classmethod
def from_gpm(
    cls,
    gpm: GenotypePhenotypeMap,
    neighbor_function: str | Callable[..., bool] = "hamming",
    cutoff: int = 1,
) -> GenotypePhenotypeGraph

Build a graph from a gpmap-v2 GenotypePhenotypeMap.

Parameters¶

gpm (GenotypePhenotypeMap, required)

A gpmap-v2 instance. Will be stashed as G.gpm.

neighbor_function (str | Callable, default "hamming")

"hamming", "codon", or a f(g1, g2, cutoff=...) -> bool callable. Determines how neighbors are detected.

cutoff (int, default 1)

Maximum neighbor distance. Must be >= 0. cutoff == 0 yields no edges.

Returns¶

A GenotypePhenotypeGraph with:

• One node per row in gpm.data, carrying all of its columns as node attributes.
• One directed edge per neighbor pair, each undirected pair appearing as both (i, j) and (j, i).

add_model¶

def add_model(
    self,
    column: str | None = None,
    model: str | Callable[..., Any] | None = None,
    **model_params: Any,
) -> None

Populate the edge attribute prob with a fixation model's output.

Parameters¶

column (str | None, default None)

Column in gpm.data to use as the node fitness. If None, every node is given fitness 1.0.

model (str | Callable | None, default None)

One of "sswm", "ratio", "moran", "mcclandish", or a callable f(fi, fj, **kwargs). None assigns 1.0 to every edge.

**model_params

Forwarded to the model on every call. The common one is population_size for moran / mcclandish.

Notes¶

The vectorized fast path uses the model's array entry point. If the supplied callable raises TypeError on arrays, add_model falls back to a Python for loop.

model¶

def model(self, v1: float, v2: float) -> float

Evaluate the last-added fixation model at a single (v1, v2) pair. Raises GpgraphError if add_model has not been called yet.

gpm (property)¶

@property
def gpm(self) -> GenotypePhenotypeMap

Returns the attached map. Raises GpgraphError if the graph was constructed without one.

read_json and read_csv¶

@classmethod
def read_json(cls, fname: str) -> GenotypePhenotypeGraph

@classmethod
def read_csv(cls, fname: str) -> GenotypePhenotypeGraph

Load a gpmap-v2 JSON or CSV save and lift it into a graph with default neighbor settings (hamming, cutoff 1). For graph-level serialization (with prob attributes), use NetworkX's own writers.

Inherited NetworkX surface¶

Because GenotypePhenotypeGraph subclasses nx.DiGraph, the full NetworkX API is available: G.nodes, G.edges, G.add_node, G.degree, nx.has_path(G, ...), etc. The class adds gpmap-specific construction and the prob edge attribute convention; it does not hide or rename any of the parent API.